feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
"""Tool: ``save_plan``. Publish sandbox Markdown for review or sharing.
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
|
feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
Reads the Markdown file the agent created in the sandbox and publishes it to the
|
|
|
|
|
plan-review page. In plan mode it is an approvable implementation plan; outside
|
|
|
|
|
plan mode it is read-only shared content.
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
"""
|
|
|
|
|
|
|
|
|
|
from __future__ import annotations
|
|
|
|
|
|
|
|
|
|
import logging
|
feat: Re-land deferred upstream features on modular webhooks (#80) (#128)
* fix(webhooks): fall back to vision model for Slack/Linear image threads
Re-land upstream #1626 onto the modular webhook structure. When a
Slack mention or Linear issue carries images but the resolved model is
text-only, fall back to a vision-capable model instead of dropping the
images. Re-points default_vision_model_pair at the fork's image-capable
models (Opus 4.8 default, else any supports_images model) rather than
upstream's openai:/anthropic: provider filter.
Refs #80, upstream #1626
* fix(slack): persist trace_message_ts so web-handoff updates the trace reply
Re-land upstream #1630 onto the modular structure. The first-mention
store_slack_run_mapping call did not pass trace_message_ts, so it was
never persisted (nothing to preserve from on first mention) and
_notify_slack_web_handoff always skipped the trace-reply update on web
handoff. Pass it through and cover it with a test.
Refs #80, upstream #1630
* feat(slack): include channel context in Slack prompts
Re-land upstream #1633 onto the modular structure. Fetch cached Slack
channel metadata once per event (_get_slack_channel_context) and thread
it through the docs-plz gate, repo resolution, and process_slack_mention
so prompts carry the channel name and a clearly-marked untrusted
channel description. Avoids duplicate conversations.info calls.
Refs #80, upstream #1633
* feat(tools): add slack_start_new_thread breakout tool
Re-land upstream #1638 onto the modular structure. Adds the
slack_start_new_thread tool (posts a top-level Slack message and
dispatches a fresh agent run for a broken-out task via the durable
dispatch_agent_run contract), wires it into the agent tool list and
tools/__init__, adds prompt guidance, and excludes it from plan mode so
it can't bypass the approval flow. Tool imports only live modules.
Refs #80, upstream #1638
* feat(plan): notify Slack on plan approval
Re-land upstream #1632 onto the modular structure. When a plan is
approved via the dashboard approve endpoint, post a thread reply to the
originating Slack thread noting the comment count and approver, after
the follow-up run is dispatched. Slack post failures never break
approval. Adapted to the fork's approve_plan (no plan_markdown read).
Refs #80, upstream #1632
* feat(plan): publish plans from sandbox files
Re-land upstream #1635 onto the modular structure, completing the
partially-ported change so dev is internally consistent. save_plan now
takes a plan_file_path, reads the agent-authored Markdown file from
/workspace/plans/ (validating extension/location/UTF-8/size) and
publishes it, instead of taking a plan_markdown string. Removes
write_file/edit_file from PLAN_MODE_EXCLUDED_TOOLS so the agent can
author the plan file, updates enter_plan_mode/reject_plan guidance and
the e2e fake LLM. Skips the #1610-only update_plan hunk (not on dev).
Refs #80, upstream #1635
* fix(security): SSRF-harden server-side image fetch + stop logging raw image URLs
INJ-01 (high): fetch_image_block used follow_redirects=True with no per-hop
revalidation and discarded the resolved-IP pin, so an attacker-authored Slack/
Linear image URL could 302-redirect the fetch to an internal host / cloud
metadata endpoint (blind SSRF), and DNS-rebinding could bypass the one-shot
is_url_safe check. Route image fetches through the same per-hop resolve+pin+
revalidate loop the http_request tool uses, lifted into url_safety as the shared
request_with_safe_redirects. Also strip the per-host Slack/Linear bearer token
on redirect so it can't be replayed to a redirect target.
SC-1 (low): linear.py logged full image URLs (which can carry signed tokens) at
DEBUG; multimodal logged them at INFO on every fetch. Log host-only.
Sink lived in multimodal.py (unchanged by the feature work) but PR #128 widened
its reach by no longer dropping images for text-only models. Fixing on the base
branch so #130/#129 inherit it on rebase. Adds fetch_image_block SSRF regression
tests (redirect-to-internal blocked; auth stripped on redirect).
2026-07-08 18:32:43 -04:00
|
|
|
from collections.abc import Mapping
|
feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
from typing import Annotated, Any
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
|
|
|
|
|
from langgraph.config import get_config
|
feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
from langgraph.prebuilt import InjectedState
|
|
|
|
|
|
|
|
|
|
from ..dashboard.plan_store import (
|
|
|
|
|
PLAN_FILE_DIRECTORY,
|
|
|
|
|
PLAN_STATUS_READY,
|
|
|
|
|
PLAN_STATUS_SHARED,
|
|
|
|
|
save_plan_content,
|
|
|
|
|
)
|
feat: Re-land deferred upstream features on modular webhooks (#80) (#128)
* fix(webhooks): fall back to vision model for Slack/Linear image threads
Re-land upstream #1626 onto the modular webhook structure. When a
Slack mention or Linear issue carries images but the resolved model is
text-only, fall back to a vision-capable model instead of dropping the
images. Re-points default_vision_model_pair at the fork's image-capable
models (Opus 4.8 default, else any supports_images model) rather than
upstream's openai:/anthropic: provider filter.
Refs #80, upstream #1626
* fix(slack): persist trace_message_ts so web-handoff updates the trace reply
Re-land upstream #1630 onto the modular structure. The first-mention
store_slack_run_mapping call did not pass trace_message_ts, so it was
never persisted (nothing to preserve from on first mention) and
_notify_slack_web_handoff always skipped the trace-reply update on web
handoff. Pass it through and cover it with a test.
Refs #80, upstream #1630
* feat(slack): include channel context in Slack prompts
Re-land upstream #1633 onto the modular structure. Fetch cached Slack
channel metadata once per event (_get_slack_channel_context) and thread
it through the docs-plz gate, repo resolution, and process_slack_mention
so prompts carry the channel name and a clearly-marked untrusted
channel description. Avoids duplicate conversations.info calls.
Refs #80, upstream #1633
* feat(tools): add slack_start_new_thread breakout tool
Re-land upstream #1638 onto the modular structure. Adds the
slack_start_new_thread tool (posts a top-level Slack message and
dispatches a fresh agent run for a broken-out task via the durable
dispatch_agent_run contract), wires it into the agent tool list and
tools/__init__, adds prompt guidance, and excludes it from plan mode so
it can't bypass the approval flow. Tool imports only live modules.
Refs #80, upstream #1638
* feat(plan): notify Slack on plan approval
Re-land upstream #1632 onto the modular structure. When a plan is
approved via the dashboard approve endpoint, post a thread reply to the
originating Slack thread noting the comment count and approver, after
the follow-up run is dispatched. Slack post failures never break
approval. Adapted to the fork's approve_plan (no plan_markdown read).
Refs #80, upstream #1632
* feat(plan): publish plans from sandbox files
Re-land upstream #1635 onto the modular structure, completing the
partially-ported change so dev is internally consistent. save_plan now
takes a plan_file_path, reads the agent-authored Markdown file from
/workspace/plans/ (validating extension/location/UTF-8/size) and
publishes it, instead of taking a plan_markdown string. Removes
write_file/edit_file from PLAN_MODE_EXCLUDED_TOOLS so the agent can
author the plan file, updates enter_plan_mode/reject_plan guidance and
the e2e fake LLM. Skips the #1610-only update_plan hunk (not on dev).
Refs #80, upstream #1635
* fix(security): SSRF-harden server-side image fetch + stop logging raw image URLs
INJ-01 (high): fetch_image_block used follow_redirects=True with no per-hop
revalidation and discarded the resolved-IP pin, so an attacker-authored Slack/
Linear image URL could 302-redirect the fetch to an internal host / cloud
metadata endpoint (blind SSRF), and DNS-rebinding could bypass the one-shot
is_url_safe check. Route image fetches through the same per-hop resolve+pin+
revalidate loop the http_request tool uses, lifted into url_safety as the shared
request_with_safe_redirects. Also strip the per-host Slack/Linear bearer token
on redirect so it can't be replayed to a redirect target.
SC-1 (low): linear.py logged full image URLs (which can carry signed tokens) at
DEBUG; multimodal logged them at INFO on every fetch. Log host-only.
Sink lived in multimodal.py (unchanged by the feature work) but PR #128 widened
its reach by no longer dropping images for text-only models. Fixing on the base
branch so #130/#129 inherit it on rebase. Adds fetch_image_block SSRF regression
tests (redirect-to-internal blocked; auth stripped on redirect).
2026-07-08 18:32:43 -04:00
|
|
|
from ..utils.sandbox_state import get_sandbox_backend
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
|
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
feat: Re-land deferred upstream features on modular webhooks (#80) (#128)
* fix(webhooks): fall back to vision model for Slack/Linear image threads
Re-land upstream #1626 onto the modular webhook structure. When a
Slack mention or Linear issue carries images but the resolved model is
text-only, fall back to a vision-capable model instead of dropping the
images. Re-points default_vision_model_pair at the fork's image-capable
models (Opus 4.8 default, else any supports_images model) rather than
upstream's openai:/anthropic: provider filter.
Refs #80, upstream #1626
* fix(slack): persist trace_message_ts so web-handoff updates the trace reply
Re-land upstream #1630 onto the modular structure. The first-mention
store_slack_run_mapping call did not pass trace_message_ts, so it was
never persisted (nothing to preserve from on first mention) and
_notify_slack_web_handoff always skipped the trace-reply update on web
handoff. Pass it through and cover it with a test.
Refs #80, upstream #1630
* feat(slack): include channel context in Slack prompts
Re-land upstream #1633 onto the modular structure. Fetch cached Slack
channel metadata once per event (_get_slack_channel_context) and thread
it through the docs-plz gate, repo resolution, and process_slack_mention
so prompts carry the channel name and a clearly-marked untrusted
channel description. Avoids duplicate conversations.info calls.
Refs #80, upstream #1633
* feat(tools): add slack_start_new_thread breakout tool
Re-land upstream #1638 onto the modular structure. Adds the
slack_start_new_thread tool (posts a top-level Slack message and
dispatches a fresh agent run for a broken-out task via the durable
dispatch_agent_run contract), wires it into the agent tool list and
tools/__init__, adds prompt guidance, and excludes it from plan mode so
it can't bypass the approval flow. Tool imports only live modules.
Refs #80, upstream #1638
* feat(plan): notify Slack on plan approval
Re-land upstream #1632 onto the modular structure. When a plan is
approved via the dashboard approve endpoint, post a thread reply to the
originating Slack thread noting the comment count and approver, after
the follow-up run is dispatched. Slack post failures never break
approval. Adapted to the fork's approve_plan (no plan_markdown read).
Refs #80, upstream #1632
* feat(plan): publish plans from sandbox files
Re-land upstream #1635 onto the modular structure, completing the
partially-ported change so dev is internally consistent. save_plan now
takes a plan_file_path, reads the agent-authored Markdown file from
/workspace/plans/ (validating extension/location/UTF-8/size) and
publishes it, instead of taking a plan_markdown string. Removes
write_file/edit_file from PLAN_MODE_EXCLUDED_TOOLS so the agent can
author the plan file, updates enter_plan_mode/reject_plan guidance and
the e2e fake LLM. Skips the #1610-only update_plan hunk (not on dev).
Refs #80, upstream #1635
* fix(security): SSRF-harden server-side image fetch + stop logging raw image URLs
INJ-01 (high): fetch_image_block used follow_redirects=True with no per-hop
revalidation and discarded the resolved-IP pin, so an attacker-authored Slack/
Linear image URL could 302-redirect the fetch to an internal host / cloud
metadata endpoint (blind SSRF), and DNS-rebinding could bypass the one-shot
is_url_safe check. Route image fetches through the same per-hop resolve+pin+
revalidate loop the http_request tool uses, lifted into url_safety as the shared
request_with_safe_redirects. Also strip the per-host Slack/Linear bearer token
on redirect so it can't be replayed to a redirect target.
SC-1 (low): linear.py logged full image URLs (which can carry signed tokens) at
DEBUG; multimodal logged them at INFO on every fetch. Log host-only.
Sink lived in multimodal.py (unchanged by the feature work) but PR #128 widened
its reach by no longer dropping images for text-only models. Fixing on the base
branch so #130/#129 inherit it on rebase. Adds fetch_image_block SSRF regression
tests (redirect-to-internal blocked; auth stripped on redirect).
2026-07-08 18:32:43 -04:00
|
|
|
_MAX_PLAN_LINES = 20_000
|
|
|
|
|
_MARKDOWN_EXTENSIONS = (".md", ".markdown")
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
|
|
|
|
|
|
feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
async def save_plan(
|
|
|
|
|
plan_file_path: str,
|
|
|
|
|
state: Annotated[dict[str, Any] | None, InjectedState] = None,
|
|
|
|
|
) -> dict[str, Any]:
|
feat: Re-land deferred upstream features on modular webhooks (#80) (#128)
* fix(webhooks): fall back to vision model for Slack/Linear image threads
Re-land upstream #1626 onto the modular webhook structure. When a
Slack mention or Linear issue carries images but the resolved model is
text-only, fall back to a vision-capable model instead of dropping the
images. Re-points default_vision_model_pair at the fork's image-capable
models (Opus 4.8 default, else any supports_images model) rather than
upstream's openai:/anthropic: provider filter.
Refs #80, upstream #1626
* fix(slack): persist trace_message_ts so web-handoff updates the trace reply
Re-land upstream #1630 onto the modular structure. The first-mention
store_slack_run_mapping call did not pass trace_message_ts, so it was
never persisted (nothing to preserve from on first mention) and
_notify_slack_web_handoff always skipped the trace-reply update on web
handoff. Pass it through and cover it with a test.
Refs #80, upstream #1630
* feat(slack): include channel context in Slack prompts
Re-land upstream #1633 onto the modular structure. Fetch cached Slack
channel metadata once per event (_get_slack_channel_context) and thread
it through the docs-plz gate, repo resolution, and process_slack_mention
so prompts carry the channel name and a clearly-marked untrusted
channel description. Avoids duplicate conversations.info calls.
Refs #80, upstream #1633
* feat(tools): add slack_start_new_thread breakout tool
Re-land upstream #1638 onto the modular structure. Adds the
slack_start_new_thread tool (posts a top-level Slack message and
dispatches a fresh agent run for a broken-out task via the durable
dispatch_agent_run contract), wires it into the agent tool list and
tools/__init__, adds prompt guidance, and excludes it from plan mode so
it can't bypass the approval flow. Tool imports only live modules.
Refs #80, upstream #1638
* feat(plan): notify Slack on plan approval
Re-land upstream #1632 onto the modular structure. When a plan is
approved via the dashboard approve endpoint, post a thread reply to the
originating Slack thread noting the comment count and approver, after
the follow-up run is dispatched. Slack post failures never break
approval. Adapted to the fork's approve_plan (no plan_markdown read).
Refs #80, upstream #1632
* feat(plan): publish plans from sandbox files
Re-land upstream #1635 onto the modular structure, completing the
partially-ported change so dev is internally consistent. save_plan now
takes a plan_file_path, reads the agent-authored Markdown file from
/workspace/plans/ (validating extension/location/UTF-8/size) and
publishes it, instead of taking a plan_markdown string. Removes
write_file/edit_file from PLAN_MODE_EXCLUDED_TOOLS so the agent can
author the plan file, updates enter_plan_mode/reject_plan guidance and
the e2e fake LLM. Skips the #1610-only update_plan hunk (not on dev).
Refs #80, upstream #1635
* fix(security): SSRF-harden server-side image fetch + stop logging raw image URLs
INJ-01 (high): fetch_image_block used follow_redirects=True with no per-hop
revalidation and discarded the resolved-IP pin, so an attacker-authored Slack/
Linear image URL could 302-redirect the fetch to an internal host / cloud
metadata endpoint (blind SSRF), and DNS-rebinding could bypass the one-shot
is_url_safe check. Route image fetches through the same per-hop resolve+pin+
revalidate loop the http_request tool uses, lifted into url_safety as the shared
request_with_safe_redirects. Also strip the per-host Slack/Linear bearer token
on redirect so it can't be replayed to a redirect target.
SC-1 (low): linear.py logged full image URLs (which can carry signed tokens) at
DEBUG; multimodal logged them at INFO on every fetch. Log host-only.
Sink lived in multimodal.py (unchanged by the feature work) but PR #128 widened
its reach by no longer dropping images for text-only models. Fixing on the base
branch so #130/#129 inherit it on rebase. Adds fetch_image_block SSRF regression
tests (redirect-to-internal blocked; auth stripped on redirect).
2026-07-08 18:32:43 -04:00
|
|
|
"""Publish a Markdown plan file from the sandbox for review.
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
|
feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
Use this in plan mode once your plan is ready. Outside plan mode, use it to
|
|
|
|
|
share long Slack responses without switching the thread into plan mode. First
|
|
|
|
|
create a Markdown file under ``/workspace/plans/`` using a dated, descriptive
|
|
|
|
|
filename, then pass that file path here. The file contents are published to
|
|
|
|
|
the plan-review page linked in the conversation. In plan mode, the user can
|
|
|
|
|
comment, approve, or request changes; outside plan mode, the page is read-only
|
|
|
|
|
shared content.
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
|
feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
Write the content in standard Markdown — headings, bullet/numbered lists, and
|
2026-07-24 18:49:53 -04:00
|
|
|
fenced code blocks all render. Shared responses persist Markdown text only;
|
|
|
|
|
they do not upload or serve sandbox-local or relative image files. If images
|
|
|
|
|
or screenshots are needed for a Slack response, post them directly in Slack
|
|
|
|
|
instead of relying on the shared response page.
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
|
|
|
|
|
Args:
|
feat: Re-land deferred upstream features on modular webhooks (#80) (#128)
* fix(webhooks): fall back to vision model for Slack/Linear image threads
Re-land upstream #1626 onto the modular webhook structure. When a
Slack mention or Linear issue carries images but the resolved model is
text-only, fall back to a vision-capable model instead of dropping the
images. Re-points default_vision_model_pair at the fork's image-capable
models (Opus 4.8 default, else any supports_images model) rather than
upstream's openai:/anthropic: provider filter.
Refs #80, upstream #1626
* fix(slack): persist trace_message_ts so web-handoff updates the trace reply
Re-land upstream #1630 onto the modular structure. The first-mention
store_slack_run_mapping call did not pass trace_message_ts, so it was
never persisted (nothing to preserve from on first mention) and
_notify_slack_web_handoff always skipped the trace-reply update on web
handoff. Pass it through and cover it with a test.
Refs #80, upstream #1630
* feat(slack): include channel context in Slack prompts
Re-land upstream #1633 onto the modular structure. Fetch cached Slack
channel metadata once per event (_get_slack_channel_context) and thread
it through the docs-plz gate, repo resolution, and process_slack_mention
so prompts carry the channel name and a clearly-marked untrusted
channel description. Avoids duplicate conversations.info calls.
Refs #80, upstream #1633
* feat(tools): add slack_start_new_thread breakout tool
Re-land upstream #1638 onto the modular structure. Adds the
slack_start_new_thread tool (posts a top-level Slack message and
dispatches a fresh agent run for a broken-out task via the durable
dispatch_agent_run contract), wires it into the agent tool list and
tools/__init__, adds prompt guidance, and excludes it from plan mode so
it can't bypass the approval flow. Tool imports only live modules.
Refs #80, upstream #1638
* feat(plan): notify Slack on plan approval
Re-land upstream #1632 onto the modular structure. When a plan is
approved via the dashboard approve endpoint, post a thread reply to the
originating Slack thread noting the comment count and approver, after
the follow-up run is dispatched. Slack post failures never break
approval. Adapted to the fork's approve_plan (no plan_markdown read).
Refs #80, upstream #1632
* feat(plan): publish plans from sandbox files
Re-land upstream #1635 onto the modular structure, completing the
partially-ported change so dev is internally consistent. save_plan now
takes a plan_file_path, reads the agent-authored Markdown file from
/workspace/plans/ (validating extension/location/UTF-8/size) and
publishes it, instead of taking a plan_markdown string. Removes
write_file/edit_file from PLAN_MODE_EXCLUDED_TOOLS so the agent can
author the plan file, updates enter_plan_mode/reject_plan guidance and
the e2e fake LLM. Skips the #1610-only update_plan hunk (not on dev).
Refs #80, upstream #1635
* fix(security): SSRF-harden server-side image fetch + stop logging raw image URLs
INJ-01 (high): fetch_image_block used follow_redirects=True with no per-hop
revalidation and discarded the resolved-IP pin, so an attacker-authored Slack/
Linear image URL could 302-redirect the fetch to an internal host / cloud
metadata endpoint (blind SSRF), and DNS-rebinding could bypass the one-shot
is_url_safe check. Route image fetches through the same per-hop resolve+pin+
revalidate loop the http_request tool uses, lifted into url_safety as the shared
request_with_safe_redirects. Also strip the per-host Slack/Linear bearer token
on redirect so it can't be replayed to a redirect target.
SC-1 (low): linear.py logged full image URLs (which can carry signed tokens) at
DEBUG; multimodal logged them at INFO on every fetch. Log host-only.
Sink lived in multimodal.py (unchanged by the feature work) but PR #128 widened
its reach by no longer dropping images for text-only models. Fixing on the base
branch so #130/#129 inherit it on rebase. Adds fetch_image_block SSRF regression
tests (redirect-to-internal blocked; auth stripped on redirect).
2026-07-08 18:32:43 -04:00
|
|
|
plan_file_path: Path to the Markdown plan file in the sandbox.
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
|
|
|
|
|
Returns:
|
|
|
|
|
``{success: True, path}`` on success, or ``{success: False, error}``.
|
|
|
|
|
"""
|
feat: Re-land deferred upstream features on modular webhooks (#80) (#128)
* fix(webhooks): fall back to vision model for Slack/Linear image threads
Re-land upstream #1626 onto the modular webhook structure. When a
Slack mention or Linear issue carries images but the resolved model is
text-only, fall back to a vision-capable model instead of dropping the
images. Re-points default_vision_model_pair at the fork's image-capable
models (Opus 4.8 default, else any supports_images model) rather than
upstream's openai:/anthropic: provider filter.
Refs #80, upstream #1626
* fix(slack): persist trace_message_ts so web-handoff updates the trace reply
Re-land upstream #1630 onto the modular structure. The first-mention
store_slack_run_mapping call did not pass trace_message_ts, so it was
never persisted (nothing to preserve from on first mention) and
_notify_slack_web_handoff always skipped the trace-reply update on web
handoff. Pass it through and cover it with a test.
Refs #80, upstream #1630
* feat(slack): include channel context in Slack prompts
Re-land upstream #1633 onto the modular structure. Fetch cached Slack
channel metadata once per event (_get_slack_channel_context) and thread
it through the docs-plz gate, repo resolution, and process_slack_mention
so prompts carry the channel name and a clearly-marked untrusted
channel description. Avoids duplicate conversations.info calls.
Refs #80, upstream #1633
* feat(tools): add slack_start_new_thread breakout tool
Re-land upstream #1638 onto the modular structure. Adds the
slack_start_new_thread tool (posts a top-level Slack message and
dispatches a fresh agent run for a broken-out task via the durable
dispatch_agent_run contract), wires it into the agent tool list and
tools/__init__, adds prompt guidance, and excludes it from plan mode so
it can't bypass the approval flow. Tool imports only live modules.
Refs #80, upstream #1638
* feat(plan): notify Slack on plan approval
Re-land upstream #1632 onto the modular structure. When a plan is
approved via the dashboard approve endpoint, post a thread reply to the
originating Slack thread noting the comment count and approver, after
the follow-up run is dispatched. Slack post failures never break
approval. Adapted to the fork's approve_plan (no plan_markdown read).
Refs #80, upstream #1632
* feat(plan): publish plans from sandbox files
Re-land upstream #1635 onto the modular structure, completing the
partially-ported change so dev is internally consistent. save_plan now
takes a plan_file_path, reads the agent-authored Markdown file from
/workspace/plans/ (validating extension/location/UTF-8/size) and
publishes it, instead of taking a plan_markdown string. Removes
write_file/edit_file from PLAN_MODE_EXCLUDED_TOOLS so the agent can
author the plan file, updates enter_plan_mode/reject_plan guidance and
the e2e fake LLM. Skips the #1610-only update_plan hunk (not on dev).
Refs #80, upstream #1635
* fix(security): SSRF-harden server-side image fetch + stop logging raw image URLs
INJ-01 (high): fetch_image_block used follow_redirects=True with no per-hop
revalidation and discarded the resolved-IP pin, so an attacker-authored Slack/
Linear image URL could 302-redirect the fetch to an internal host / cloud
metadata endpoint (blind SSRF), and DNS-rebinding could bypass the one-shot
is_url_safe check. Route image fetches through the same per-hop resolve+pin+
revalidate loop the http_request tool uses, lifted into url_safety as the shared
request_with_safe_redirects. Also strip the per-host Slack/Linear bearer token
on redirect so it can't be replayed to a redirect target.
SC-1 (low): linear.py logged full image URLs (which can carry signed tokens) at
DEBUG; multimodal logged them at INFO on every fetch. Log host-only.
Sink lived in multimodal.py (unchanged by the feature work) but PR #128 widened
its reach by no longer dropping images for text-only models. Fixing on the base
branch so #130/#129 inherit it on rebase. Adds fetch_image_block SSRF regression
tests (redirect-to-internal blocked; auth stripped on redirect).
2026-07-08 18:32:43 -04:00
|
|
|
if not isinstance(plan_file_path, str):
|
|
|
|
|
return {"success": False, "error": "plan_file_path must be a string"}
|
|
|
|
|
path = plan_file_path.strip()
|
|
|
|
|
if not path:
|
|
|
|
|
return {"success": False, "error": "plan_file_path cannot be empty"}
|
|
|
|
|
if not _is_markdown_path(path):
|
|
|
|
|
return {
|
|
|
|
|
"success": False,
|
|
|
|
|
"error": f"plan_file_path must point to a Markdown file in {PLAN_FILE_DIRECTORY}",
|
|
|
|
|
}
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
|
|
|
|
|
try:
|
|
|
|
|
config = get_config()
|
|
|
|
|
except Exception:
|
|
|
|
|
config = {}
|
|
|
|
|
configurable = config.get("configurable", {}) if isinstance(config, dict) else {}
|
|
|
|
|
thread_id = configurable.get("thread_id") if isinstance(configurable, dict) else None
|
|
|
|
|
if not thread_id:
|
|
|
|
|
return {"success": False, "error": "no thread_id in run config"}
|
|
|
|
|
|
|
|
|
|
try:
|
feat: Re-land deferred upstream features on modular webhooks (#80) (#128)
* fix(webhooks): fall back to vision model for Slack/Linear image threads
Re-land upstream #1626 onto the modular webhook structure. When a
Slack mention or Linear issue carries images but the resolved model is
text-only, fall back to a vision-capable model instead of dropping the
images. Re-points default_vision_model_pair at the fork's image-capable
models (Opus 4.8 default, else any supports_images model) rather than
upstream's openai:/anthropic: provider filter.
Refs #80, upstream #1626
* fix(slack): persist trace_message_ts so web-handoff updates the trace reply
Re-land upstream #1630 onto the modular structure. The first-mention
store_slack_run_mapping call did not pass trace_message_ts, so it was
never persisted (nothing to preserve from on first mention) and
_notify_slack_web_handoff always skipped the trace-reply update on web
handoff. Pass it through and cover it with a test.
Refs #80, upstream #1630
* feat(slack): include channel context in Slack prompts
Re-land upstream #1633 onto the modular structure. Fetch cached Slack
channel metadata once per event (_get_slack_channel_context) and thread
it through the docs-plz gate, repo resolution, and process_slack_mention
so prompts carry the channel name and a clearly-marked untrusted
channel description. Avoids duplicate conversations.info calls.
Refs #80, upstream #1633
* feat(tools): add slack_start_new_thread breakout tool
Re-land upstream #1638 onto the modular structure. Adds the
slack_start_new_thread tool (posts a top-level Slack message and
dispatches a fresh agent run for a broken-out task via the durable
dispatch_agent_run contract), wires it into the agent tool list and
tools/__init__, adds prompt guidance, and excludes it from plan mode so
it can't bypass the approval flow. Tool imports only live modules.
Refs #80, upstream #1638
* feat(plan): notify Slack on plan approval
Re-land upstream #1632 onto the modular structure. When a plan is
approved via the dashboard approve endpoint, post a thread reply to the
originating Slack thread noting the comment count and approver, after
the follow-up run is dispatched. Slack post failures never break
approval. Adapted to the fork's approve_plan (no plan_markdown read).
Refs #80, upstream #1632
* feat(plan): publish plans from sandbox files
Re-land upstream #1635 onto the modular structure, completing the
partially-ported change so dev is internally consistent. save_plan now
takes a plan_file_path, reads the agent-authored Markdown file from
/workspace/plans/ (validating extension/location/UTF-8/size) and
publishes it, instead of taking a plan_markdown string. Removes
write_file/edit_file from PLAN_MODE_EXCLUDED_TOOLS so the agent can
author the plan file, updates enter_plan_mode/reject_plan guidance and
the e2e fake LLM. Skips the #1610-only update_plan hunk (not on dev).
Refs #80, upstream #1635
* fix(security): SSRF-harden server-side image fetch + stop logging raw image URLs
INJ-01 (high): fetch_image_block used follow_redirects=True with no per-hop
revalidation and discarded the resolved-IP pin, so an attacker-authored Slack/
Linear image URL could 302-redirect the fetch to an internal host / cloud
metadata endpoint (blind SSRF), and DNS-rebinding could bypass the one-shot
is_url_safe check. Route image fetches through the same per-hop resolve+pin+
revalidate loop the http_request tool uses, lifted into url_safety as the shared
request_with_safe_redirects. Also strip the per-host Slack/Linear bearer token
on redirect so it can't be replayed to a redirect target.
SC-1 (low): linear.py logged full image URLs (which can carry signed tokens) at
DEBUG; multimodal logged them at INFO on every fetch. Log host-only.
Sink lived in multimodal.py (unchanged by the feature work) but PR #128 widened
its reach by no longer dropping images for text-only models. Fixing on the base
branch so #130/#129 inherit it on rebase. Adds fetch_image_block SSRF regression
tests (redirect-to-internal blocked; auth stripped on redirect).
2026-07-08 18:32:43 -04:00
|
|
|
content = (await _read_plan_file(str(thread_id), path)).strip()
|
|
|
|
|
if not content:
|
|
|
|
|
return {"success": False, "error": "plan file cannot be empty"}
|
feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
await _save(str(thread_id), content, path, plan_mode=_active_plan_mode(state, configurable))
|
feat: plan mode with model-driven entry and collaborative review (#1580)
* feat: add plan mode for read-only research and planning
Adds a per-run plan_mode flag that puts the agent in a read-only
research phase: a strong prompt section is injected and mutating tools
are stripped via ExcludeToolsMiddleware so the agent proposes a
reviewable implementation plan before any edits. Surfaced in the
dashboard UI with a Plan toggle (Shift+Tab) wired through the thread API.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: enforce plan-mode read-only at tool layer and disable subagents
Addresses PR review: plan mode previously relied on prompt text to keep
the shell read-only and left the task subagent (built with its own
write/PR/Linear tools) unrestricted. Now `task` is excluded so research
cannot be delegated to a mutating subagent, and a new
PlanModeShellGuardMiddleware enforces a read-only command allowlist on
`execute`, blocking writes, git state changes, installs, redirection,
and command substitution regardless of model/prompt-injection compliance.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: harden plan-mode shell guard against wrapped mutations
Block git global options that take values (-C, --git-dir, ...) from being
misread as the subcommand, reject config-injection options (-c,
--config-env, --exec-path), and drop the env command wrapper that could
run arbitrary commands.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add plan mode with enter_plan_mode tool, profile/team defaults, Slack commands and approval flow
- enter_plan_mode tool: agent self-activates plan mode via Command(update={'plan_mode': True})
- Plan mode resolution: per-thread > profile default > team default > False
- PLAN_MODE_GUIDANCE_SECTION: always-present prompt section telling agent about the tool
- profile_plan_mode_default and team plan_mode_default settings
- Slack plan on/off/status commands with thread metadata persistence
- slack_thread_reply plan_approval=True renders Approve/Revise/Cancel buttons
- Interactivity handler: approve triggers implementation run, cancel posts confirmation
- Frontend: plan_mode_default in Profile/ProfileUpdate/TeamSettings types and UI toggles
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* test: add tests for enter_plan_mode tool, profile/team defaults, Slack plan commands, approval blocks
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* refactor(plan-mode): drop shell guard, rely on prompt for read-only discipline
Remove PlanModeShellGuardMiddleware and its enforcement of read-only shell
commands during plan mode. Plan mode now relies on the system prompt to
instruct the agent not to run mutating commands; the mutating-tool exclusion
(ExcludeToolsMiddleware) is retained.
* test(open-swe): add Playwright E2E for the Slack → PR → web handoff
Local, secrets-free end-to-end suite that drives the full happy path through mock Slack/GitHub control panels and the real dashboard UI. Only the LLM and external SaaS HTTP boundaries (GitHub/Slack APIs, OAuth token mint) are faked — the real process_slack_mention, get_agent, deepagents loop, tools, middleware, and dashboard authorization all run under `langgraph dev` with a scripted fake chat model and a local temp-dir sandbox.
- full_flow: a Slack mention runs the agent, which implements a change in the sandbox, opens a PR against a fake GitHub remote, and replies with the PR link in the same thread.
- dashboard: clicking the bot's real "Open in Web" link loads the built ui/ app (served same-origin); the thread owner can continue the conversation, while a different user sees the same thread read-only (no composer).
Wired into Agent CI as a `Playwright E2E` job that runs on pull requests.
* fix(open-swe): serve E2E UI assets via explicit route; pin Playwright
The dashboard E2E served the built ui/ SPA's /assets via app.mount(StaticFiles), but LangGraph's custom-app loader serves APIRoutes and drops sub-app Mounts, so /assets 404'd under `langgraph dev` in CI — the React app never booted and the composer/transcript never rendered. Serve assets via an explicit route instead.
Also pin @playwright/test to the latest (1.61.0) for reproducible runs, and make the owner composer assertion tolerant of either hydration state.
* test(open-swe): record Playwright trace + video on every E2E run
Capture a replayable trace (DOM snapshots, network, console, source) and a screen recording for every test, not just retries, plus a screenshot on failure. The CI job already uploads playwright-report/ and test-results/, so each run now has a downloadable replay; documented how to open it.
* feat(plan-mode): collaborative plan review with BlockNote + Yjs
When the agent enters plan mode it writes the plan as a markdown file in the
sandbox (save_plan tool), publishes it, and posts a review link to the source
channel. Reviewers open the plan inside the dashboard (under the /agents shell),
read it rendered in a BlockNote editor, and leave inline comments synced live
over Yjs. Only the thread owner can approve; any reviewer can request changes.
On approve/reject the comments are harvested and handed to the agent for the
follow-up run; the agent never sees comments mid-review.
- agent: enter_plan_mode persists plan state; new save_plan tool; prompt shares
the plan-review link.
- dashboard: Yjs WebSocket collab server (pycrdt-websocket) with store-backed
snapshots; plan content/status store; plan REST API (get/approve/reject,
owner-only approve, client-harvested comments); planStatus on thread summaries.
- ui: BlockNote native comments (CommentsExtension + YjsThreadStore) plan page
mounted under the agents shell, with a "Review plan" banner in the thread view
and a back-link; theme-aware (dark mode) using the dashboard tokens.
- e2e: Playwright coverage of the full Slack -> plan -> review -> approve -> PR
flow, including cross-user comment sync and owner-only approval.
* fix(plan-mode): address review feedback (authz, overrides, leaks, deps)
- plan-collab WS: authorize per-thread before joining a room (same read gate as
the REST API) — previously any logged-in user could join any thread (IDOR).
- plan-collab: tie the snapshot flusher to active connections (refcount) so each
opened plan no longer leaks a permanent 1.5s task on the shared event loop.
- plan decisions: include thread_id in the follow-up run configurable so the run
resumes the existing thread; set plan_mode explicitly so approve forces it off.
- get_agent: an explicit per-thread plan_mode (Slack `plan off`, approved plan,
dashboard toggle) now overrides profile/team defaults instead of falling back.
- plan mode tool gating moved to a state-aware PlanModeMiddleware installed
unconditionally, so a mid-run enter_plan_mode restricts the next model turn;
before_agent resets stale plan_mode so a later run isn't forced back into it.
- exclude write-capable http_request from plan mode.
- pin pycrdt / pycrdt-websocket with upper bounds.
Includes the latest base (#1583): E2E UI assets served via explicit route
(fixes the Playwright CI failure — LangGraph's app loader drops sub-app mounts).
* style: ruff format plan_collab.py
* fix(plan-mode): owner-gate Slack approval + same-origin check on collab WS
- Slack "Approve & Implement" now verifies the clicking user is the plan
requester (owner, via the stored triggering_user_id) before implementing —
matching the dashboard API's owner-only approval. Non-owners are pointed to
Revise / feedback.
- The plan-collab WebSocket validates the handshake Origin against the dashboard
allowlist before accept() (no-op when unconfigured, e.g. local/dev), mirroring
the REST require_same_origin CSRF defense.
* fix(plan-mode): enter plan mode only via the model + local mock dev harness
Plan mode is now entered solely when the model calls enter_plan_mode.
Removed the per-user and team plan_mode_default settings (backend + UI)
and the Slack `plan on/off/status` toggle.
- enter_plan_mode returns a terminating ToolMessage, fixing the missing
ToolMessage error that silently dropped plan mode mid-run.
- PlanReview: defer Yjs provider/doc teardown so React StrictMode's dev
remount doesn't destroy and then reuse the collaboration provider.
- e2e plan_review spec asserts plan_mode actually engages.
- LangSmith trace-url resolution is best-effort: bail before any API
call when the tenant is unset, cache failures, log at debug.
- Add `pnpm run dev:mock`: same-origin Vite HMR harness with a real LLM,
Alice/Bob mock users, and a GitHub login picker.
* docs(plan-mode): drop stale references to removed profile/team defaults
The plan_mode middleware docstring and the approve/reject dispatch comment
still described the profile/team plan_mode_default resolution that no longer
exists; reword to match model-driven entry + the per-thread carry.
* feat(plan-mode): let any reviewer edit the plan, not just comment
Drop the owner/commenter split for the plan document: everyone with read
access edits and comments alike (DefaultThreadStoreAuth "editor" for all,
editor always editable until a decision, anyone seeds the empty doc). This
matches the collab WS, which already relays frames to every readable user.
Plan approval stays owner-gated.
* test(plan-mode): assert plan-mode entry via the tool's success message
plan_mode lives only in run state for tool gating; it is not a persisted
thread-state channel, so the previous `values.plan_mode === true` poll
could never pass. Assert instead that enter_plan_mode's success ToolMessage
("Plan mode is active …") lands in the thread — which only happens when the
tool's Command applies cleanly, the exact regression this guards.
---------
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
2026-06-23 15:06:58 -04:00
|
|
|
except Exception as exc: # noqa: BLE001
|
|
|
|
|
logger.exception("save_plan failed for thread %s", thread_id)
|
|
|
|
|
return {"success": False, "error": f"failed to save plan: {exc}"}
|
|
|
|
|
return {"success": True, "path": path}
|
|
|
|
|
|
|
|
|
|
|
feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
async def _save(thread_id: str, content: str, path: str, *, plan_mode: bool) -> None:
|
feat: Re-land deferred upstream features on modular webhooks (#80) (#128)
* fix(webhooks): fall back to vision model for Slack/Linear image threads
Re-land upstream #1626 onto the modular webhook structure. When a
Slack mention or Linear issue carries images but the resolved model is
text-only, fall back to a vision-capable model instead of dropping the
images. Re-points default_vision_model_pair at the fork's image-capable
models (Opus 4.8 default, else any supports_images model) rather than
upstream's openai:/anthropic: provider filter.
Refs #80, upstream #1626
* fix(slack): persist trace_message_ts so web-handoff updates the trace reply
Re-land upstream #1630 onto the modular structure. The first-mention
store_slack_run_mapping call did not pass trace_message_ts, so it was
never persisted (nothing to preserve from on first mention) and
_notify_slack_web_handoff always skipped the trace-reply update on web
handoff. Pass it through and cover it with a test.
Refs #80, upstream #1630
* feat(slack): include channel context in Slack prompts
Re-land upstream #1633 onto the modular structure. Fetch cached Slack
channel metadata once per event (_get_slack_channel_context) and thread
it through the docs-plz gate, repo resolution, and process_slack_mention
so prompts carry the channel name and a clearly-marked untrusted
channel description. Avoids duplicate conversations.info calls.
Refs #80, upstream #1633
* feat(tools): add slack_start_new_thread breakout tool
Re-land upstream #1638 onto the modular structure. Adds the
slack_start_new_thread tool (posts a top-level Slack message and
dispatches a fresh agent run for a broken-out task via the durable
dispatch_agent_run contract), wires it into the agent tool list and
tools/__init__, adds prompt guidance, and excludes it from plan mode so
it can't bypass the approval flow. Tool imports only live modules.
Refs #80, upstream #1638
* feat(plan): notify Slack on plan approval
Re-land upstream #1632 onto the modular structure. When a plan is
approved via the dashboard approve endpoint, post a thread reply to the
originating Slack thread noting the comment count and approver, after
the follow-up run is dispatched. Slack post failures never break
approval. Adapted to the fork's approve_plan (no plan_markdown read).
Refs #80, upstream #1632
* feat(plan): publish plans from sandbox files
Re-land upstream #1635 onto the modular structure, completing the
partially-ported change so dev is internally consistent. save_plan now
takes a plan_file_path, reads the agent-authored Markdown file from
/workspace/plans/ (validating extension/location/UTF-8/size) and
publishes it, instead of taking a plan_markdown string. Removes
write_file/edit_file from PLAN_MODE_EXCLUDED_TOOLS so the agent can
author the plan file, updates enter_plan_mode/reject_plan guidance and
the e2e fake LLM. Skips the #1610-only update_plan hunk (not on dev).
Refs #80, upstream #1635
* fix(security): SSRF-harden server-side image fetch + stop logging raw image URLs
INJ-01 (high): fetch_image_block used follow_redirects=True with no per-hop
revalidation and discarded the resolved-IP pin, so an attacker-authored Slack/
Linear image URL could 302-redirect the fetch to an internal host / cloud
metadata endpoint (blind SSRF), and DNS-rebinding could bypass the one-shot
is_url_safe check. Route image fetches through the same per-hop resolve+pin+
revalidate loop the http_request tool uses, lifted into url_safety as the shared
request_with_safe_redirects. Also strip the per-host Slack/Linear bearer token
on redirect so it can't be replayed to a redirect target.
SC-1 (low): linear.py logged full image URLs (which can carry signed tokens) at
DEBUG; multimodal logged them at INFO on every fetch. Log host-only.
Sink lived in multimodal.py (unchanged by the feature work) but PR #128 widened
its reach by no longer dropping images for text-only models. Fixing on the base
branch so #130/#129 inherit it on rebase. Adds fetch_image_block SSRF regression
tests (redirect-to-internal blocked; auth stripped on redirect).
2026-07-08 18:32:43 -04:00
|
|
|
await save_plan_content(
|
feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
thread_id,
|
|
|
|
|
markdown=content,
|
|
|
|
|
status=PLAN_STATUS_READY if plan_mode else PLAN_STATUS_SHARED,
|
|
|
|
|
plan_file_path=path,
|
|
|
|
|
plan_mode=plan_mode or None,
|
feat: Re-land deferred upstream features on modular webhooks (#80) (#128)
* fix(webhooks): fall back to vision model for Slack/Linear image threads
Re-land upstream #1626 onto the modular webhook structure. When a
Slack mention or Linear issue carries images but the resolved model is
text-only, fall back to a vision-capable model instead of dropping the
images. Re-points default_vision_model_pair at the fork's image-capable
models (Opus 4.8 default, else any supports_images model) rather than
upstream's openai:/anthropic: provider filter.
Refs #80, upstream #1626
* fix(slack): persist trace_message_ts so web-handoff updates the trace reply
Re-land upstream #1630 onto the modular structure. The first-mention
store_slack_run_mapping call did not pass trace_message_ts, so it was
never persisted (nothing to preserve from on first mention) and
_notify_slack_web_handoff always skipped the trace-reply update on web
handoff. Pass it through and cover it with a test.
Refs #80, upstream #1630
* feat(slack): include channel context in Slack prompts
Re-land upstream #1633 onto the modular structure. Fetch cached Slack
channel metadata once per event (_get_slack_channel_context) and thread
it through the docs-plz gate, repo resolution, and process_slack_mention
so prompts carry the channel name and a clearly-marked untrusted
channel description. Avoids duplicate conversations.info calls.
Refs #80, upstream #1633
* feat(tools): add slack_start_new_thread breakout tool
Re-land upstream #1638 onto the modular structure. Adds the
slack_start_new_thread tool (posts a top-level Slack message and
dispatches a fresh agent run for a broken-out task via the durable
dispatch_agent_run contract), wires it into the agent tool list and
tools/__init__, adds prompt guidance, and excludes it from plan mode so
it can't bypass the approval flow. Tool imports only live modules.
Refs #80, upstream #1638
* feat(plan): notify Slack on plan approval
Re-land upstream #1632 onto the modular structure. When a plan is
approved via the dashboard approve endpoint, post a thread reply to the
originating Slack thread noting the comment count and approver, after
the follow-up run is dispatched. Slack post failures never break
approval. Adapted to the fork's approve_plan (no plan_markdown read).
Refs #80, upstream #1632
* feat(plan): publish plans from sandbox files
Re-land upstream #1635 onto the modular structure, completing the
partially-ported change so dev is internally consistent. save_plan now
takes a plan_file_path, reads the agent-authored Markdown file from
/workspace/plans/ (validating extension/location/UTF-8/size) and
publishes it, instead of taking a plan_markdown string. Removes
write_file/edit_file from PLAN_MODE_EXCLUDED_TOOLS so the agent can
author the plan file, updates enter_plan_mode/reject_plan guidance and
the e2e fake LLM. Skips the #1610-only update_plan hunk (not on dev).
Refs #80, upstream #1635
* fix(security): SSRF-harden server-side image fetch + stop logging raw image URLs
INJ-01 (high): fetch_image_block used follow_redirects=True with no per-hop
revalidation and discarded the resolved-IP pin, so an attacker-authored Slack/
Linear image URL could 302-redirect the fetch to an internal host / cloud
metadata endpoint (blind SSRF), and DNS-rebinding could bypass the one-shot
is_url_safe check. Route image fetches through the same per-hop resolve+pin+
revalidate loop the http_request tool uses, lifted into url_safety as the shared
request_with_safe_redirects. Also strip the per-host Slack/Linear bearer token
on redirect so it can't be replayed to a redirect target.
SC-1 (low): linear.py logged full image URLs (which can carry signed tokens) at
DEBUG; multimodal logged them at INFO on every fetch. Log host-only.
Sink lived in multimodal.py (unchanged by the feature work) but PR #128 widened
its reach by no longer dropping images for text-only models. Fixing on the base
branch so #130/#129 inherit it on rebase. Adds fetch_image_block SSRF regression
tests (redirect-to-internal blocked; auth stripped on redirect).
2026-07-08 18:32:43 -04:00
|
|
|
)
|
|
|
|
|
|
|
|
|
|
|
feat: port plan-review & workflow-approval UX (#159)
* feat: port plan-review & workflow-approval UX (#135)
Port six upstream commits onto dev:
- c03a6be7 (already ported): keep plan guidance high-level
- 546042a4: add workflow approval UI with diff preview, approval URLs,
web review links, and polling for approval status during active runs
- 216cf181: remove workflow token elevation; approved pushes pass
through directly without proxy token rewriting
- 3dbc0282: preserve plan redirects after login by accepting relative
same-origin redirect_to values and rejecting blocked paths
- bb104d93: submit plan comments with cmd+enter
- 90cb6caa: terse Slack replies, shared content via save_plan outside
plan mode (PLAN_STATUS_SHARED), reject shared-content mutations
Refs: #135
* fix: restore login page render and clear CI lint/format
The plan-review port removed the authRedirectUrl import from login.tsx
but left its call site, crashing the login page at runtime (blank page,
no 'Sign in to open-swe'). Pass the relative path straight to loginUrl,
matching the plan route and the backend relative-redirect handling.
Also drop an unused os import in the guard test and reformat
workflow_push_guard.py to satisfy ruff.
* fix: carry workflows:write on the standing proxy token
Complete the half-ported upstream 216cf181 cascade. The port dropped
_run_with_workflow_token from the guard but missed the paired github_app
change, so an approved .github/workflows push ran with the base token
(no workflows:write) and GitHub 403'd it.
Add workflows:write to BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS and delete the
now-orphaned WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant; update the
github_app and proxy_auth tests to match. The HITL approval gate in
workflow_push_guard.py is unchanged — this only lets the standing token
push once a human approves.
* fix: restore transient workflow-token elevation (revert standing workflows:write)
The standing GitHub-App proxy token (BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS) is
ALWAYS-ON, so carrying workflows:write on it made the fork's HITL workflow-push
guard the sole control over unapproved workflow pushes. The guard's git-push
parser has gaps (obfuscated-expansion push, `gh api` REST contents PUT,
fully-qualified cross-branch refspecs); with a permanently workflows-scoped
token those gaps become live unapproved-workflow-push exploits (1 critical, 2
high — security review BLOCK on #159).
Restore dev's transient-elevation model:
- Drop workflows:write from BASE_RUNTIME_PROXY_TOKEN_PERMISSIONS; re-add the
WORKFLOW_RUNTIME_PROXY_TOKEN_PERMISSIONS constant (base + workflows:write).
- Re-introduce _run_with_workflow_token in the guard: it mints the
workflows-scoped token via refresh_proxy_token around the approved,
guard-normalized fixed_command, then downscopes to RUNTIME then BASE in a
finally. Route the approval branch through it.
- Restore the dev token/elevation tests.
The standing token no longer carries workflows:write, so the three parser
bypasses hit GitHub 403 again; an approved push still succeeds because the
elevation grants workflows:write only around the normalized command. Keeps all
of #159's diff-preview / approval-URL / Slack-card guard additions.
* fix: reject protocol-relative path from sanitizeAuthRedirect (open redirect)
sanitizeAuthRedirect returned parsed.pathname+search+hash, which `new URL` can
resolve to a protocol-relative `//host` (e.g. input `/..//evil.com` normalizes
same-origin, passing the origin check, but yields a path starting with `//`).
ClientRedirect / login.tsx feed that path to window.location.replace, so it
navigates cross-origin — an open redirect. Reject any resolved path that is not
a single-leading-slash path (`^/[^/]`), falling back to the default. Adds
coverage for `/..//evil.com`, `/.//evil.com`, and `//evil.com`.
* fix: log SECURITY error when workflow-token downscope fails
The elevate->push->downscope finally block was silent on failure. If both
refresh_proxy_token calls fail, the sandbox retains workflows:write for the
rest of the run with no signal. Log a SECURITY error on the partial and full
downscope-failure paths so the retention is observable.
Addresses the GPT-4.1 cross-family review of the token-scope remediation.
---------
Co-authored-by: amoussa1229 <166072409+amoussa1229@users.noreply.github.com>
Co-authored-by: Adam Moussa <adam@seahavenind.com>
2026-07-09 16:03:13 -04:00
|
|
|
def _active_plan_mode(state: dict[str, Any] | None, configurable: Any) -> bool:
|
|
|
|
|
if isinstance(state, dict) and state.get("plan_mode") is True:
|
|
|
|
|
return True
|
|
|
|
|
return isinstance(configurable, dict) and configurable.get("plan_mode") is True
|
|
|
|
|
|
|
|
|
|
|
feat: Re-land deferred upstream features on modular webhooks (#80) (#128)
* fix(webhooks): fall back to vision model for Slack/Linear image threads
Re-land upstream #1626 onto the modular webhook structure. When a
Slack mention or Linear issue carries images but the resolved model is
text-only, fall back to a vision-capable model instead of dropping the
images. Re-points default_vision_model_pair at the fork's image-capable
models (Opus 4.8 default, else any supports_images model) rather than
upstream's openai:/anthropic: provider filter.
Refs #80, upstream #1626
* fix(slack): persist trace_message_ts so web-handoff updates the trace reply
Re-land upstream #1630 onto the modular structure. The first-mention
store_slack_run_mapping call did not pass trace_message_ts, so it was
never persisted (nothing to preserve from on first mention) and
_notify_slack_web_handoff always skipped the trace-reply update on web
handoff. Pass it through and cover it with a test.
Refs #80, upstream #1630
* feat(slack): include channel context in Slack prompts
Re-land upstream #1633 onto the modular structure. Fetch cached Slack
channel metadata once per event (_get_slack_channel_context) and thread
it through the docs-plz gate, repo resolution, and process_slack_mention
so prompts carry the channel name and a clearly-marked untrusted
channel description. Avoids duplicate conversations.info calls.
Refs #80, upstream #1633
* feat(tools): add slack_start_new_thread breakout tool
Re-land upstream #1638 onto the modular structure. Adds the
slack_start_new_thread tool (posts a top-level Slack message and
dispatches a fresh agent run for a broken-out task via the durable
dispatch_agent_run contract), wires it into the agent tool list and
tools/__init__, adds prompt guidance, and excludes it from plan mode so
it can't bypass the approval flow. Tool imports only live modules.
Refs #80, upstream #1638
* feat(plan): notify Slack on plan approval
Re-land upstream #1632 onto the modular structure. When a plan is
approved via the dashboard approve endpoint, post a thread reply to the
originating Slack thread noting the comment count and approver, after
the follow-up run is dispatched. Slack post failures never break
approval. Adapted to the fork's approve_plan (no plan_markdown read).
Refs #80, upstream #1632
* feat(plan): publish plans from sandbox files
Re-land upstream #1635 onto the modular structure, completing the
partially-ported change so dev is internally consistent. save_plan now
takes a plan_file_path, reads the agent-authored Markdown file from
/workspace/plans/ (validating extension/location/UTF-8/size) and
publishes it, instead of taking a plan_markdown string. Removes
write_file/edit_file from PLAN_MODE_EXCLUDED_TOOLS so the agent can
author the plan file, updates enter_plan_mode/reject_plan guidance and
the e2e fake LLM. Skips the #1610-only update_plan hunk (not on dev).
Refs #80, upstream #1635
* fix(security): SSRF-harden server-side image fetch + stop logging raw image URLs
INJ-01 (high): fetch_image_block used follow_redirects=True with no per-hop
revalidation and discarded the resolved-IP pin, so an attacker-authored Slack/
Linear image URL could 302-redirect the fetch to an internal host / cloud
metadata endpoint (blind SSRF), and DNS-rebinding could bypass the one-shot
is_url_safe check. Route image fetches through the same per-hop resolve+pin+
revalidate loop the http_request tool uses, lifted into url_safety as the shared
request_with_safe_redirects. Also strip the per-host Slack/Linear bearer token
on redirect so it can't be replayed to a redirect target.
SC-1 (low): linear.py logged full image URLs (which can carry signed tokens) at
DEBUG; multimodal logged them at INFO on every fetch. Log host-only.
Sink lived in multimodal.py (unchanged by the feature work) but PR #128 widened
its reach by no longer dropping images for text-only models. Fixing on the base
branch so #130/#129 inherit it on rebase. Adds fetch_image_block SSRF regression
tests (redirect-to-internal blocked; auth stripped on redirect).
2026-07-08 18:32:43 -04:00
|
|
|
async def _read_plan_file(thread_id: str, path: str) -> str:
|
|
|
|
|
backend = await get_sandbox_backend(thread_id)
|
|
|
|
|
result = await backend.aread(path, offset=0, limit=_MAX_PLAN_LINES)
|
|
|
|
|
error = _value(result, "error")
|
|
|
|
|
if error:
|
|
|
|
|
raise ValueError(error)
|
|
|
|
|
file_data = _value(result, "file_data")
|
|
|
|
|
if file_data is None:
|
|
|
|
|
raise ValueError("plan file could not be read")
|
|
|
|
|
encoding = _value(file_data, "encoding")
|
|
|
|
|
if encoding is not None and encoding != "utf-8":
|
|
|
|
|
raise ValueError("plan file must be UTF-8 text")
|
|
|
|
|
content = _value(file_data, "content")
|
|
|
|
|
if not isinstance(content, str):
|
|
|
|
|
raise ValueError("plan file content was not text")
|
|
|
|
|
if content.count("\n") + 1 >= _MAX_PLAN_LINES:
|
|
|
|
|
raise ValueError("plan file is too large")
|
|
|
|
|
return content
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def _value(value: Any, key: str) -> Any:
|
|
|
|
|
if isinstance(value, Mapping):
|
|
|
|
|
return value.get(key)
|
|
|
|
|
return getattr(value, key, None)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def _is_markdown_path(path: str) -> bool:
|
|
|
|
|
if "\x00" in path or not path.startswith(f"{PLAN_FILE_DIRECTORY}/"):
|
|
|
|
|
return False
|
|
|
|
|
filename = path.removeprefix(f"{PLAN_FILE_DIRECTORY}/")
|
|
|
|
|
if not filename or "/" in filename:
|
|
|
|
|
return False
|
|
|
|
|
return filename.lower().endswith(_MARKDOWN_EXTENSIONS)
|