open-swe/agent/middleware/refresh_slack_status.py

232 lines
8.2 KiB
Python
Raw Permalink Normal View History

feat: add optional Slack Assistants API typing status indicator (#1269) * 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>
2026-05-08 10:21:55 -07:00
"""Middleware that keeps Slack's assistant status current during agent work.
Slack's ``assistant.threads.setStatus`` indicator expires after two minutes
feat: add optional Slack Assistants API typing status indicator (#1269) * 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>
2026-05-08 10:21:55 -07:00
if no message is sent. This middleware refreshes the indicator while model and
tool calls are actively running, then clears it when the run exits without
relying on the model to post a final Slack reply.
"""
from __future__ import annotations
import asyncio
import contextlib
import logging
from collections.abc import Awaitable, Callable
from typing import Any, TypeVar
from langchain.agents.middleware.types import (
AgentMiddleware,
AgentState,
ModelRequest,
ModelResponse,
)
from langchain_core.messages import ToolMessage
from langgraph.config import get_config
from langgraph.prebuilt.tool_node import ToolCallRequest
from langgraph.runtime import Runtime
from langgraph.types import Command
from ..utils.slack import (
DEFAULT_ASSISTANT_STATUS,
DEFAULT_LOADING_MESSAGES,
set_slack_assistant_status,
)
logger = logging.getLogger(__name__)
_HEARTBEAT_INTERVAL_SECONDS = 60.0
_MAX_HEARTBEAT_SECONDS = 60 * 60
_T = TypeVar("_T")
# Tool-name -> human-readable status. Keep in sync with the tool list in
# agent/server.py, agent/reviewer.py, and the deepagents built-ins (read_file,
# write_file, edit_file, execute, glob, grep, task).
_TOOL_STATUS: dict[str, str] = {
"read_file": "reading files...",
"write_file": "editing files...",
"edit_file": "editing files...",
"execute": "running commands...",
"glob": "scanning the repo...",
"grep": "searching the codebase...",
"task": "delegating to a subagent...",
"web_search": "searching the web...",
"fetch_url": "fetching a URL...",
"http_request": "making an HTTP request...",
"request_pr_review": "requesting a PR review...",
"slack_add_reaction": "reacting in Slack...",
feat: add optional Slack Assistants API typing status indicator (#1269) * 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>
2026-05-08 10:21:55 -07:00
"slack_read_thread_messages": "reading Slack history...",
"slack_thread_reply": "drafting a Slack reply...",
"linear_comment": "commenting on Linear...",
"linear_create_issue": "creating a Linear issue...",
"linear_get_issue": "checking Linear...",
"linear_get_issue_comments": "checking Linear...",
"linear_list_teams": "checking Linear...",
"linear_search_issues": "searching Linear...",
feat: add optional Slack Assistants API typing status indicator (#1269) * 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>
2026-05-08 10:21:55 -07:00
"linear_update_issue": "updating Linear...",
"linear_delete_issue": "updating Linear...",
"add_finding": "recording review findings...",
"update_finding": "updating review findings...",
"list_findings": "checking review findings...",
"publish_review": "publishing the review...",
}
def _slack_thread_from_config() -> tuple[str, str] | None:
config = get_config()
configurable = config.get("configurable", {}) if isinstance(config, dict) else {}
slack_thread = configurable.get("slack_thread") if isinstance(configurable, dict) else None
if not isinstance(slack_thread, dict):
return None
channel_id = slack_thread.get("channel_id")
thread_ts = slack_thread.get("thread_ts")
if not isinstance(channel_id, str) or not isinstance(thread_ts, str):
return None
if not channel_id or not thread_ts:
return None
return channel_id, thread_ts
def _tool_call_name(tool_call: object) -> str | None:
if isinstance(tool_call, dict):
name = tool_call.get("name")
else:
name = getattr(tool_call, "name", None)
return name if isinstance(name, str) and name else None
def _status_from_recent_tool_calls(messages: list[Any]) -> str:
"""Pick a status string based on the last assistant message's tool calls."""
for msg in reversed(messages):
tool_calls = getattr(msg, "tool_calls", None)
if not tool_calls:
continue
# Use the first tool call's name; if the agent fans out, this is fine
# as a single-line indicator.
name = _tool_call_name(tool_calls[0])
if isinstance(name, str) and name in _TOOL_STATUS:
return _TOOL_STATUS[name]
return DEFAULT_ASSISTANT_STATUS
return DEFAULT_ASSISTANT_STATUS
async def _set_status(channel_id: str, thread_ts: str, status: str) -> None:
try:
await set_slack_assistant_status(
channel_id,
thread_ts,
status=status,
loading_messages=list(DEFAULT_LOADING_MESSAGES) if status else None,
)
except Exception:
logger.exception("Failed to update Slack assistant status")
class SlackAssistantStatusMiddleware(AgentMiddleware):
"""Maintain Slack's assistant status for Slack-triggered agent runs."""
state_schema = AgentState
def __init__(
self,
*,
heartbeat_interval_seconds: float = _HEARTBEAT_INTERVAL_SECONDS,
max_heartbeat_seconds: float = _MAX_HEARTBEAT_SECONDS,
) -> None:
self._heartbeat_interval_seconds = heartbeat_interval_seconds
self._max_heartbeat_seconds = max_heartbeat_seconds
async def abefore_agent(
self,
state: AgentState, # noqa: ARG002
runtime: Runtime, # noqa: ARG002
) -> dict[str, Any] | None:
await self._try_set(DEFAULT_ASSISTANT_STATUS)
return None
async def aafter_agent(
self,
state: AgentState, # noqa: ARG002
runtime: Runtime, # noqa: ARG002
) -> dict[str, Any] | None:
# Slack auto-clears on bot replies. This explicit clear covers soft
# exits where the agent stops without posting a message.
await self._try_set("")
return None
async def awrap_model_call(
self,
request: ModelRequest,
handler: Callable[[ModelRequest], Awaitable[ModelResponse]],
) -> ModelResponse:
messages = request.state.get("messages", []) if isinstance(request.state, dict) else []
status = _status_from_recent_tool_calls(messages)
return await self._run_with_heartbeat(status, handler(request))
async def awrap_tool_call(
self,
request: ToolCallRequest,
handler: Callable[[ToolCallRequest], Awaitable[ToolMessage | Command]],
) -> ToolMessage | Command:
name = _tool_call_name(request.tool_call)
status = _TOOL_STATUS.get(name or "", DEFAULT_ASSISTANT_STATUS)
return await self._run_with_heartbeat(status, handler(request))
def wrap_model_call(
self,
request: ModelRequest,
handler: Callable[[ModelRequest], ModelResponse],
) -> ModelResponse:
return handler(request)
def wrap_tool_call(
self,
request: ToolCallRequest,
handler: Callable[[ToolCallRequest], ToolMessage | Command],
) -> ToolMessage | Command:
return handler(request)
async def _try_set(self, status: str) -> None:
try:
slack_thread = _slack_thread_from_config()
if slack_thread is None:
return
channel_id, thread_ts = slack_thread
await _set_status(channel_id, thread_ts, status)
except Exception:
logger.exception("Failed to read Slack thread config")
async def _run_with_heartbeat(self, status: str, awaitable: Awaitable[_T]) -> _T:
try:
slack_thread = _slack_thread_from_config()
except Exception:
logger.exception("Failed to read Slack thread config")
return await awaitable
if slack_thread is None:
return await awaitable
channel_id, thread_ts = slack_thread
await _set_status(channel_id, thread_ts, status)
heartbeat = asyncio.create_task(self._heartbeat(channel_id, thread_ts, status))
try:
return await awaitable
finally:
heartbeat.cancel()
with contextlib.suppress(asyncio.CancelledError):
await heartbeat
async def _heartbeat(self, channel_id: str, thread_ts: str, status: str) -> None:
loop = asyncio.get_running_loop()
started_at = loop.time()
while True:
await asyncio.sleep(self._heartbeat_interval_seconds)
if loop.time() - started_at >= self._max_heartbeat_seconds:
logger.info(
"Stopping Slack assistant status heartbeat after %.0f seconds",
self._max_heartbeat_seconds,
)
return
await _set_status(channel_id, thread_ts, status)