* feat: tighten reviewer eval workflow
Require the reviewer to verify and dedupe findings before recording them, and make benchmark runs safe to execute against deployed reviewer graphs without posting GitHub reviews.
* chore: move reviewer eval settings to config
Load reviewer benchmark settings from the default eval config file so deployed eval runs do not require a wide CLI surface.
* feat: allow reviewer eval model overrides
Pass reviewer model and reasoning effort from the eval config into reviewer runs so isolated benchmark deployments can test Opus 4.7 high thinking.
* fix: use adaptive thinking for Opus 4.7
Switch Opus 4.7 model overrides to Anthropic adaptive thinking with effort instead of the deprecated budgeted thinking payload rejected by the API.
* refactor: use latest Anthropic effort API
Remove legacy Anthropic budget-token thinking support and route Anthropic efforts through adaptive thinking plus effort.
* revert prompting
* feat: dashboard backend — GitHub OAuth, profile CRUD, admin endpoints
Adds agent/dashboard/ FastAPI router mounted at /dashboard/api covering:
- GitHub App OAuth login → JWT cookie session (cross-domain ready)
- profile CRUD against LangGraph Store with model+effort validation
- admin gate via CONFIGURED_ADMINS
- /repos via /user/installations using the user's encrypted OAuth token
CORS allowlist on webapp.py is opt-in via DASHBOARD_ALLOWED_ORIGINS so the
Vercel-hosted frontend can call the LangSmith deployment with credentials.
* feat: apply dashboard profile model/effort overrides in get_agent
Look up the triggering user's GitHub login from config (direct field or
GITHUB_USER_EMAIL_MAP reverse lookup), read their profile from the Store,
and apply default_model + reasoning_effort to make_model when both are
valid. Effort 'max' is captured on the profile but not yet wired through —
the OpenAI Reasoning Literal doesn't accept it.
* feat: ui/ TanStack Start dashboard for profile config
Scaffolded with the shadcn b7CScJIjA preset (TanStack Start template,
base-ui primitives, Tailwind v4). Three routes:
- /login — Sign in with GitHub (links to /dashboard/api/auth/login)
- /profile — Edit default model, reasoning effort, default repo
- /admin — Admin-only: list users and edit other profiles
API client (src/lib/api.ts) uses credentials: include so the osw_session
cookie set by the OAuth callback rides cross-origin. VITE_DASHBOARD_API_BASE_URL
points at the LangSmith deployment.
Effort options re-render when the model changes; 'max' on Opus 4.7 is
captured on the profile but ignored downstream until anthropic reasoning
is wired through make_model.
* feat: searchable Combobox for default repo picker
Replaces the Select with a base-ui Combobox so users can filter by typing,
the popup is wider than the trigger so full owner/repo names are readable,
and the list caps at max-h-80 to stay on screen.
* fix: address review comments + wire default_repo and Anthropic thinking
Security/correctness fixes from PR review:
* Open redirect: validate `redirect_to` in `/auth/login` against
`DASHBOARD_BASE_URL` + `DASHBOARD_ALLOWED_ORIGINS` before signing it
into the state JWT. Anything off-allowlist falls back to the dashboard
base URL. (PR #1302 r3250054386)
* Login CSRF: bind the OAuth `state` to the requesting browser. At
`/auth/login` we generate a fresh nonce, set it as a short-lived
HttpOnly SameSite=Lax cookie scoped to `/dashboard/api/auth`, and
embed `hash_state_nonce(nonce)` in the state JWT. At `/auth/callback`
we require the cookie nonce to hash-match the state JWT's nonce_hash
(constant-time compare). (PR #1302 r3250054395)
* RMW race in profile vs token writes: split storage into two
namespaces — `["profiles"]` for user-editable settings and
`["oauth_tokens"]` for the encrypted GitHub token. Each upsert now
only writes its own namespace so an in-flight profile save can no
longer clobber a fresh token from a concurrent re-login (and vice
versa). (PR #1302 r3250054393)
* /repos pagination: follow `Link: rel="next"` for both
`/user/installations` and per-installation `/repositories` with
per_page=100, capped at 1000 items. (PR #1302 r3250054401)
Feature wires:
* default_repo: applied as a fallback in `get_slack_repo_config` (after
explicit-repo / thread metadata, before the env defaults) and in the
Linear webhook (after comment-body extraction, before team mapping).
Both paths resolve the triggering user's GitHub login via
GITHUB_USER_EMAIL_MAP and read the profile's default_repo.
* Anthropic "thinking" effort: `make_model` now accepts a `thinking`
kwarg; `get_agent` maps profile effort {low,medium,high,xhigh,max}
to budget_tokens {1k,4k,12k,32k,60k} when the chosen model is
anthropic. OpenAI path still ignores "max" since the Literal doesn't
accept it.
* feat: TTL and revocation handling for cached GitHub OAuth tokens [closes AB-2322]
Persist github_token_expires_at alongside github_token_encrypted, treat
expired cache entries as missing so we re-resolve before kicking off
runs, and invalidate the cached ciphertext on a downstream 401 so the
next invocation gets a fresh token instead of replaying a revoked one.
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
* webapp: forward installation-token expiry to reviewer cache writes
The three reviewer-thread persist sites in webapp.py were calling
get_github_app_installation_token() (no expiry) and persist_encrypted_github_token
without expires_at, so cached App tokens were treated as never-expiring even
though they actually expire in ~1 hour.
---------
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
Co-authored-by: Johannes du Plessis <51395795+johannes117@users.noreply.github.com>
Co-authored-by: Johannes du Plessis <johannes@langchain.dev>
When the primary model raises a transient provider error (5xx, 429,
connection/timeout) the request is retried once against a fallback
model from the other provider. Anthropic primaries fall back to
OpenAI and vice versa. Also bumps the SDK max_retries from the
default 2 to 6 so quick blips stay on the primary and keep prompt
caching warm.
Triggered by 529 OverloadedError traces that ended runs silently
with no Slack/Linear/PR reply.
* fix: recover from mid-run sandbox death
Recreate dead sandboxes during tool execution and stop repeated unrecoverable timeout loops with a user-facing notification.
* fix: count repeated sandbox recreations
Treat consecutive sandbox recreations as an unrecovered failure streak so outages cannot loop until the model-call limit.
The gh-cli migration removed agent/middleware/open_pr.py and
agent/tools/commit_and_open_pr.py — the only callers of
add_user_coauthor_trailer / add_pr_collaboration_note. Since then the
agent has been driving commits and PRs entirely via gh, with no
attribution back to the Slack/Linear/GitHub user who triggered the run.
Resolve the triggering user's identity in get_agent (reusing the
existing authorship helpers) and inject a Collaborative Attribution
section into the system prompt with the exact trailer and PR-body note
to use. The section is only rendered when an identity is resolvable, so
runs without a known triggering user are unchanged.
* feat: add optional Slack Assistants API typing status indicator
Mirrors OpenClaw's pragmatic approach: instead of rebuilding around
assistant_thread_started events, just opt into assistants.threads.setStatus
to show 'is thinking…' while the agent is working, and clear it when
post_slack_thread_reply lands. Gated behind SLACK_ASSISTANTS_API_ENABLED so
it can be toggled without touching code.
* fix(slack): drop redundant clear, add status heartbeat across model calls
- Slack auto-clears the typing indicator on bot post; remove the explicit
assistants.threads.setStatus("") call from post_slack_thread_reply.
- The indicator expires after ~2 minutes; add a before_model middleware
that refreshes it on every model tick so it stays visible across long
agent runs. Reuses the existing slack_thread.{channel_id,thread_ts}
configurable already plumbed for notify_step_limit.
- chat:write is sufficient on the bot token (assistant:write is on the
way out per Slack docs); no scope or app-config change required.
* feat(slack): contextual status text + rotating loading_messages
- set_slack_assistant_status now accepts an optional loading_messages list
(capped at 10 per Slack's API), surfaced via the assistants.threads.setStatus
payload so Slack rotates through them client-side.
- The heartbeat middleware derives a contextual status from the last
assistant message's tool calls (e.g. "searching the codebase…" after
grep, "running commands…" after execute), falling back to the default
"is thinking…" when no tool calls or unknown tool name.
- Adds a curated DEFAULT_LOADING_MESSAGES list passed alongside the
contextual status on each refresh.
* fix slack assistant status lifecycle
---------
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
Co-authored-by: Johannes du Plessis <johannes@langchain.dev>
* feat: add reviewer graph + eval target wiring
- New `reviewer` graph (`agent/reviewer.py`) registered in langgraph.json
alongside the main `agent` graph. Reuses the same sandbox lifecycle,
GH proxy auth, and middleware primitives from `agent.server`, but with
a narrower tool set, a reviewer-specific system prompt, no
commit/push, and the `task` (subagent) tool stripped via
`_ToolExclusionMiddleware` so review stays in one context.
- New `github_comment` tool: agents call it once per issue with
`(file, line, body, severity)` and the eval scores those calls
against golden comments.
- `ensure_no_empty_msg` middleware (the no_op nudge) is intentionally
*not* on the reviewer's stack — that middleware exists to enforce the
main agent's "always finalize via Slack/Linear/PR" contract, which
the reviewer doesn't have. The main agent's behavior is unchanged.
- `evals/reviewer/target.py`: send PR info as a user message, extract
every `github_comment` tool call (multiple expected per review) into
the run output.
- `evals/reviewer/judge.py`: per-example evaluator now returns a list
of metrics under `{"results": [...]}` so LangSmith averages each
numeric key (f1/precision/recall/tp/fp/fn) across the experiment in
the UI. Dropped the broken `aggregate_pr` summary evaluator that
reached for an attribute that doesn't exist on `RunTree`.
- `evals/reviewer/run_eval.py`: `--limit` now slices the dataset via
`client.list_examples(limit=N)` since `aevaluate` doesn't accept
`max_examples`.
- Makefile: `dev` and `run` targets now use `uv run` so they work
without an activated venv.
* resolve comments
---------
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: re-apply git identity on every agent invocation, source from env
Cached/reconnected sandboxes weren't getting `git config --global` re-applied
(the call was gated on new-sandbox creation), so commits picked up whatever
identity the sandbox happened to have — sometimes an email not associated
with any GitHub account, which Vercel rejects on preview deploys.
Move the git config call out of the new-sandbox branch so it runs on every
get_agent, and source the values from OPEN_SWE_GIT_AUTHOR_NAME /
OPEN_SWE_GIT_AUTHOR_EMAIL env vars (defaulting to the existing bot identity)
so forks can override without code changes.
* cleanup
---------
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: move github workflows to gh cli
Use LangSmith proxy auth to support gh-driven GitHub workflows while removing custom GitHub wrapper tools.
* docker ignore + snapshot and docker image updates
* updated image and instructions
* removing open_pr if needed after agent call
* feat: add edit_pull_request tool for editing PR titles and descriptions
* fix: patch auth flow in open PR middleware tests
* fix: support app token for editing PRs
---------
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
Co-authored-by: Johannes du Plessis <johannes@langchain.dev>
- Root cause: LLM occasionally generates strings like '1, 80' or '170, "limit": 60'
for integer fields, causing a Pydantic ValidationError and wasting an LLM turn
- Change: add SanitizeToolInputsMiddleware in agent/middleware/sanitize_tool_inputs.py
that extracts the leading integer from any string value in offset/limit before
the call reaches Pydantic validation; registered before ToolErrorMiddleware in server.py
- Verified: 14 unit tests covering all three production trace patterns pass
Co-authored-by: LangSmith Forge <forge-agent@langsmith.ai>
Co-authored-by: Johannes du Plessis <johannes@langchain.dev>
* fix: notify users via Slack when agent hits model call step limit
- Root cause: GraphRecursionError at 1000 steps bypassed all @after_agent
middleware including open_pr_if_needed, leaving users with no notification
- Change: Added ModelCallLimitMiddleware(run_limit=60) to intercept gracefully
before the hard recursion limit, and added notify_step_limit_reached
@after_agent middleware to post a Slack thread reply when the limit fires
- Verified: 107 existing tests pass, no regressions
* fix: harden step-limit Slack notification
Ensure the step-limit notification runs after the PR safety net and cover the new middleware behavior with focused unit tests.
---------
Co-authored-by: LangSmith Forge <forge-agent@langsmith.ai>
Co-authored-by: Johannes du Plessis <johannes@langchain.dev>
* feat: add github CI check run tools for shepherding CI
Add get_pr_check_runs and rerun_failed_check_runs tools that authenticate
using the GitHub App installation token so the agent can query and retry
CI status on private repos without relying on GH_TOKEN or unauthenticated
http_request calls.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: handle paginated GitHub CI results
* refactor(github_ci): address review feedback
- Rename rerun_failed_check_runs -> rerun_failed_workflow_runs and clarify
in docstrings that the tool only retries GitHub Actions workflow runs
(not third-party CI checks surfaced by get_pr_check_runs).
- Skip action_required workflow runs when rerunning; those need manual
approval, not a rerun.
- Run rerun-failed-jobs requests concurrently via asyncio.gather instead
of sequentially.
- Fix latent pagination bug in _fetch_paginated_items where caller-supplied
params could overwrite per_page/page and break the end-of-pagination
check; reserved keys now always win and the threshold uses a PER_PAGE
constant.
- Set an explicit 30s httpx timeout so a hung GitHub call cannot stall
the agent loop.
- Restore alphabetical ordering of tools in agent/tools/__init__.py.
- Add tests for: a 500 surfaced on a later pagination page, and
action_required runs being filtered out of rerun candidates.
---------
Co-authored-by: Claude Agent <agent@anthropic.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Johannes du Plessis <johannes@langchain.dev>
* feat: add get_pr_review_comments tool for authenticated GitHub API access
The agent was asking users to paste PR review comments because it had no
tool to fetch them with auth. This adds get_pr_review_comments, which uses
the GitHub App installation token to fetch all three comment types (thread
comments, inline review comments, review submissions) from private repos.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* u
* u
---------
Co-authored-by: Forge Agent <agent@forge.ai>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Palash Shah <palash@langchain.dev>
Co-authored-by: Palash Shah <35114859+Palashio@users.noreply.github.com>
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: default to GPT-5.5 medium reasoning
Use OpenAI GPT-5.5 with medium reasoning as the default model and document the completion-token budget semantics for reasoning models.
* fix: use Responses API reasoning config
Pass GPT-5.5 reasoning settings through LangChain's Responses API parameter instead of the Chat Completions-only reasoning_effort field.
* feat: raise GPT-5.5 output budget
Set the default GPT-5.5 output token budget to the model maximum so long-running coding tasks have more room for reasoning and final responses.
* feat: align recursion limit with Deep Agents
Use Deep Agents' default recursion limit so longer coding runs have room to complete without Open SWE imposing a lower cap.
* chore: remove minimal effort level
* chore: reduce max tokens to 64_000
---------
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* fix: add Exa web search tool
* chore: update uv.lock for exa-py dependency
* linting
* chore: remove web_search from system prompt
* chore: drop search_type and category params from web_search
* Add GitHub PR review tools (list, get, create, update, dismiss, submit, list comments) and bind them to the agent
* format n lint
---------
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
Co-authored-by: Aran Yogesh <yogesh.mahendran@langchain.dev>
Adds 6 new agent tools backed by Linear's GraphQL API, with a shared
_graphql_request helper to reduce boilerplate. Refactors existing
comment_on_linear_issue to use the same helper.
Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
* feat: add GitHub PR comment trigger and reply support
* refactor: improve readability of GitHub integration
* linting
* fix: resolve github token from thread metadata and improve PR trigger flow
* fix: fall back to OAuth for GitHub webhook when no token in thread metadata
* give me commit message github integeration working without a breaking
* auth.py refactor
* fix: validate cached GitHub token before use to handle expiry
* liniting
* ci unitest formatting
* feat: post PR comments as GitHub App bot instead of user OAuth token
* resolved comments
* slack resolveed comments
* Refactor docstring and comments in get_slack_repo_config
Removed unnecessary comments and cleaned up docstring formatting.
* cr
* cr
* cr
* cr
---------
Co-authored-by: bracesproul <braceasproul@gmail.com>
