* fix(todo-middleware): prevent premature agent exit with incomplete todos
When plan mode is active (is_plan_mode=True), the agent occasionally
exits the loop and outputs a final response while todo items are still
incomplete. This happens because the routing edge only checks for
tool_calls, not todo completion state.
Fixes#2112
Add an after_model override to TodoMiddleware with
@hook_config(can_jump_to=["model"]). When the model produces a
response with no tool calls but there are still incomplete todos, the
middleware injects a todo_completion_reminder HumanMessage and returns
jump_to=model to force another model turn. A cap of 2 reminders
prevents infinite loops when the agent cannot make further progress.
Also adds _completion_reminder_count() helper and 14 new unit tests
covering all edge cases of the new after_model / aafter_model logic.
* Remove unnecessary blank line in test file
* Fix runtime argument annotation in before_model
* Apply suggestions from code review
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
---------
Co-authored-by: octo-patch <octo-patch@github.com>
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* docs: mark memory updater async migration as completed
- Update TODO.md to mark the replacement of sync model.invoke()
with async model.ainvoke() in title_middleware and memory updater
as completed using [x] format
Addresses #2131
* feat: switch memory updater to async LLM calls
- Add async aupdate_memory() method using await model.ainvoke()
- Convert sync update_memory() to use async wrapper
- Add _run_async_update_sync() for nested loop context handling
- Maintain backward compatibility with existing sync API
- Add ThreadPoolExecutor for async execution from sync contexts
Addresses #2131
* test: add tests for async memory updater
- Add test_async_update_memory_uses_ainvoke() to verify async path
- Convert existing tests to use AsyncMock and ainvoke assertions
- Add test_sync_update_memory_wrapper_works_in_running_loop()
- Update all model mocks to use async await patterns
Addresses #2131
* fix: apply ruff formatting to memory updater
- Format multi-line expressions to single line
- Ensure code style consistency with project standards
- Fix lint issues caught by GitHub Actions
* test: add comprehensive tests for async memory updater
- Add test_async_update_memory_uses_ainvoke() to verify async path
- Convert existing tests to use AsyncMock and ainvoke assertions
- Add test_sync_update_memory_wrapper_works_in_running_loop()
- Update all model mocks to use async await patterns
- Ensure backward compatibility with sync API
* fix: satisfy ruff formatting in memory updater test
---------
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
* fix uploads for mounted sandbox providers
* Potential fix for pull request finding
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
---------
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
* fix: resolve Windows pnpm detection in check script
* style: format check script regression test
* Potential fix for pull request finding
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
* fix: resolve corepack fallback on windows
---------
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
* fix(title): strip <think> tags from title model responses and assistant context
Reasoning models (e.g. minimax M2.7, DeepSeek-R1) emit <think>...</think>
blocks before their actual output. When such a model is used as the title
model (or as the main agent), the raw thinking content leaked into the thread
title stored in state, so the chat list showed the internal monologue instead
of a meaningful title.
Fixes#1884
- Add `_strip_think_tags()` helper using a regex to remove all <think>...</think> blocks
- Apply it in `_parse_title()` so the title model response is always clean
- Apply it to the assistant message in `_build_title_prompt()` so thinking
content from the first AI turn is not fed back to the title model
- Add four new unit tests covering: stripping in parse, think-only response,
assistant prompt stripping, and end-to-end async flow with think tags
* Fix the lint error
---------
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
* fix: disable custom-agent management API by default
* style: format agents API hardening files
* fix: address review feedback for agents API hardening
* fix: add missing disabled API coverage
* fix: wrap blocking readability call with asyncio.to_thread in web_fetch
The readability extractor internally spawns a Node.js subprocess via
readabilipy, which blocks the async event loop and causes a
BlockingError when web_fetch is invoked inside LangGraph's async
runtime.
Wrap the synchronous extract_article call with asyncio.to_thread to
offload it to a thread pool, unblocking the event loop.
Note: community/infoquest/tools.py has the same latent issue and
should be addressed in a follow-up PR.
Closes#2152
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* test: verify web_fetch offloads extraction via asyncio.to_thread
Add a regression test that monkeypatches asyncio.to_thread to confirm
readability extraction is offloaded to a worker thread, preventing
future refactors from reintroducing the blocking call.
Addresses Copilot review feedback on #2157.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
* feat: set up Vitest frontend testing infrastructure with CI workflow
Migrate existing 4 frontend test files from Node.js native test runner
(node:test + node:assert/strict) to Vitest, reorganize test directory
structure under tests/unit/ mirroring src/ layout, and add a dedicated
CI workflow for frontend unit tests.
- Add vitest as devDependency, remove tsx
- Create vitest.config.ts with @/ path alias
- Migrate tests to Vitest API (test/expect/vi)
- Rename .mjs test files to .ts
- Move tests from src/ to tests/unit/ (mirrors src/ layout)
- Add frontend/Makefile `test` target
- Add .github/workflows/frontend-unit-tests.yml (parallel to backend)
- Update CONTRIBUTING.md, README.md, AGENTS.md, CLAUDE.md
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* style: fix the lint error
* style: fix the lint error
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
- Move time.sleep() -> asyncio.sleep() from Planned to Completed Features
- Clean up duplicate entries in TODO.md
Ensures completed async optimizations are properly tracked.
* feat(subagents): allow model override per subagent in config.yaml
Wire the existing SubagentConfig.model field to config.yaml so users
can assign different models to different subagent types.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* test(subagents): cover model override in SubagentsAppConfig + registry
Addresses review feedback on #2064:
- registry.py: update stale inline comment — the block now applies
timeout, max_turns AND model overrides, not just timeout.
- test_subagent_timeout_config.py: add coverage for model override
resolution across SubagentOverrideConfig, SubagentsAppConfig
(get_model_for + load), and registry.get_subagent_config:
- per-agent model override is applied to registry-returned config
- omitted `model` keeps the builtin value
- explicit `model: null` in config.yaml is equivalent to omission
- model override on one agent does not affect other agents
- model override preserves all other fields (name, description,
timeout_seconds, max_turns)
- model override does not mutate BUILTIN_SUBAGENTS
Copilot's suggestion (3) "setting model to 'inherit' forces inheritance"
is skipped intentionally: there is no 'inherit' sentinel in the current
implementation — model is `str | None`, and None already means
"inherit from parent". Adding a sentinel would be a new feature, not
test coverage for this PR.
Tests run locally: 51 passed (37 existing + 14 new / expanded).
* test(subagents): reject empty-string model at config load time
Addresses WillemJiang's review comment on #2064 (empty-string edge case):
- subagents_config.py: add `min_length=1` to the `model` field on
SubagentOverrideConfig. `model: ""` in config.yaml would otherwise
bypass the `is not None` check and reach create_chat_model(name="")
as a confusing runtime error. This is symmetric with the existing
`ge=1` guards on timeout_seconds / max_turns, so the validation style
stays consistent across all three override fields.
- test_subagent_timeout_config.py: add test_rejects_empty_model
mirroring the existing test_rejects_zero / test_rejects_negative
cases; update the docstring on test_model_accepts_any_string (now
test_model_accepts_any_non_empty_string) to reflect the new guard.
Not addressing the first comment (validating `model` against the
`models:` section at load time) in this PR. `SubagentsAppConfig` is
scoped to the `subagents:` block and cannot see the sibling `models:`
section, so proper cross-section validation needs a second pass or a
structural change that is out of scope here — and the current behavior
is consistent with how timeout_seconds / max_turns work today. Happy to
track this as a follow-up issue covering cross-section validation
uniformly for all three fields.
Tests run locally: 52 passed in this file; 1847 passed, 18 skipped
across the full backend suite. Ruff check + format clean.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat(channels): add Discord channel integration
Add a Discord bot channel following the existing Telegram/Slack pattern.
The bot listens for messages, creates conversation threads, and relays
responses back to Discord with 2000-char message splitting.
- DiscordChannel extends Channel base class
- Lazy imports discord.py with install hint
- Thread-based conversations (each Discord thread maps to a DeerFlow thread)
- Allowed guilds filter for access control
- File attachment support via discord.File
- Registered in service.py and manager.py
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(channels): address Copilot review suggestions for Discord integration
- Disable @everyone/@here mentions via AllowedMentions.none()
- Add 10s timeout to client close to prevent shutdown hangs
- Log publish_inbound errors via future callback instead of silently dropping
- Open file handle on caller thread to avoid cross-thread ownership issues
- Notify user in channel when thread creation fails
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(discord): resolve lint errors in Discord channel
- Replace asyncio.TimeoutError with builtin TimeoutError (UP041)
- Remove extraneous f-string prefix (F541)
- Apply ruff format
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(tests): remove fake langgraph_sdk shim from test_discord_channel
The module-level sys.modules.setdefault shim installed a fake
langgraph_sdk.errors.ConflictError during pytest collection. Because
pytest imports all test modules before running them, test_channels.py
then imported the fake ConflictError instead of the real one.
In test_handle_feishu_stream_conflict_sends_busy_message, the test
constructs ConflictError(message, response=..., body=...). The fake
only subclasses Exception (which takes no kwargs), so the construction
raised TypeError. The manager's _is_thread_busy_error check then saw a
TypeError instead of a ConflictError and fell through to the generic
'An error occurred' message.
langgraph_sdk is a real dependency, so the shim is unnecessary.
Removing it makes both test files import the same real ConflictError
and the full suite pass (1773 passed, 15 skipped).
---------
Co-authored-by: Matt Van Horn <455140+mvanhorn@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
* fix(sandbox): resolve paths in read_file/write_file content for LocalSandbox
In LocalSandbox mode, read_file and write_file now transform
container paths in file content, matching the path handling
behavior of bash tool.
- write_file: resolves virtual paths in content to system paths
before writing, so scripts with /mnt/user-data paths work
when executed
- read_file: reverse-resolves system paths back to virtual
paths in returned content for consistency
This fixes scenarios where agents write Python scripts with
virtual paths, then execute them via bash tool expecting the
paths to work.
Fixes#1778
* fix(sandbox): address Copilot review — dedicated content resolver + forward-slash safety + tests
- Extract _resolve_paths_in_content() separate from _resolve_paths_in_command()
to decouple file-content path resolution from shell-command parsing
- Normalize resolved paths to forward slashes to avoid Windows backslash
escape issues in source files (e.g. \U in Python string literals)
- Add 4 focused tests: write resolves content, forward-slash guarantee,
read reverse-resolves content, and write→read roundtrip
* style: fix ruff lint — remove extraneous f-string prefix
* fix(sandbox): only reverse-resolve paths in agent-written files
read_file previously applied _reverse_resolve_paths_in_output to ALL
file content, which could silently rewrite paths in user uploads and
external tool output (Willem Jiang review on #1935).
Now tracks files written through write_file in _agent_written_paths.
Only those files get reverse-resolved on read. Non-agent files are
returned as-is.
---------
Co-authored-by: JasonOA888 <JasonOA888@users.noreply.github.com>
* fix(middleware): add per-tool-type frequency detection to LoopDetectionMiddleware
The existing hash-based loop detection only catches identical tool call
sets. When the agent calls the same tool type (e.g. read_file) on many
different files, each call produces a unique hash and bypasses detection.
This causes the agent to exhaust recursion_limit, consuming 150K-225K
tokens per failed run.
Add a second detection layer that tracks cumulative call counts per tool
type per thread. Warns at 30 calls (configurable) and forces stop at 50.
The hard stop message now uses the actual returned message instead of a
hardcoded constant, so both hash-based and frequency-based stops produce
accurate diagnostics.
Also fix _apply() to use the warning message returned by
_track_and_check() for hard stops, instead of always using _HARD_STOP_MSG.
Closes#1987
* Apply suggestions from code review
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* fix(lint): remove unused imports and fix line length
- Remove unused _TOOL_FREQ_HARD_STOP_MSG and _TOOL_FREQ_WARNING_MSG
imports from test file (F401)
- Break long _TOOL_FREQ_WARNING_MSG string to fit within 240 char limit (E501)
* style: apply ruff format
* test: add LRU eviction and per-thread reset coverage for frequency state
Address review feedback from @WillemJiang:
- Verify _tool_freq and _tool_freq_warned are cleaned on LRU eviction
- Add test for reset(thread_id=...) clearing only the target thread's
frequency state while leaving others intact
* fix(makefile): route Windows shell-script targets through Git Bash (#2060)
---------
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Asish Kumar <87874775+officialasishkumar@users.noreply.github.com>
* fix: improve sandbox security and preserve multimodal content
* Add unit test modifications for test_injects_uploaded_files_tag_into_list_content
* format updated_content
* Add regression tests for multimodal upload content and host bash default safety
* feat(provisioner): add optional PVC support for sandbox volumes (#1978)
Add SKILLS_PVC_NAME and USERDATA_PVC_NAME env vars to allow sandbox
Pods to use PersistentVolumeClaims instead of hostPath volumes. This
prevents data loss in production when pods are rescheduled across nodes.
When USERDATA_PVC_NAME is set, a subPath of threads/{thread_id}/user-data
is used so a single PVC can serve multiple threads. Falls back to hostPath
when the new env vars are not set, preserving backward compatibility.
* add unit test for provisioner pvc volumes
* refactor: extract shared provisioner_module fixture to conftest.py
Agent-Logs-Url: https://github.com/bytedance/deer-flow/sessions/e7ccf708-c6ba-40e4-844a-b526bdb249dd
Co-authored-by: WillemJiang <219644+WillemJiang@users.noreply.github.com>
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: JeffJiang <for-eleven@hotmail.com>
* feat(blog): implement blog structure with post listing and tagging functionality
* feat(blog): enhance blog layout and post metadata display with new components
* fix(blog): address PR #1962 review feedback and fix lint issues (#14)
* fix: format
---------
Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com>
* fix(backend): stream DeerFlowClient AI text as token deltas (#1969)
DeerFlowClient.stream() subscribed to LangGraph stream_mode=["values",
"custom"] which only delivers full-state snapshots at graph-node
boundaries, so AI replies were dumped as a single messages-tuple event
per node instead of streaming token-by-token. `client.stream("hello")`
looked identical to `client.chat("hello")` — the bug reported in #1969.
Subscribe to "messages" mode as well, forward AIMessageChunk deltas as
messages-tuple events with delta semantics (consumers accumulate by id),
and dedup the values-snapshot path so it does not re-synthesize AI
text that was already streamed. Introduce a per-id usage_metadata
counter so the final AIMessage in the values snapshot and the final
"messages" chunk — which carry the same cumulative usage — are not
double-counted.
chat() now accumulates per-id deltas and returns the last message's
full accumulated text. Non-streaming mock sources (single event per id)
are a degenerate case of the same logic, keeping existing callers and
tests backward compatible.
Verified end-to-end against a real LLM: a 15-number count emits 35
messages-tuple events with BPE subword boundaries clearly visible
("eleven" -> "ele" / "ven", "twelve" -> "tw" / "elve"), 476ms across
the window, end-event usage matches the values-snapshot usage exactly
(not doubled). tests/test_client_live.py::TestLiveStreaming passes.
New unit tests:
- test_messages_mode_emits_token_deltas: 3 AIMessageChunks produce 3
delta events with correct content/id/usage, values-snapshot does not
duplicate, usage counted once.
- test_chat_accumulates_streamed_deltas: chat() rebuilds full text
from deltas.
- test_messages_mode_tool_message: ToolMessage delivered via messages
mode is not duplicated by the values-snapshot synthesis path.
The stream() docstring now documents why this client does not reuse
Gateway's run_agent() / StreamBridge pipeline (sync vs async, raw
LangChain objects vs serialized dicts, single caller vs HTTP fan-out).
Fixes#1969
* refactor(backend): simplify DeerFlowClient streaming helpers (#1969)
Post-review cleanup for the token-level streaming fix. No behavior
change for correct inputs; one efficiency regression fixed.
Fix: chat() O(n²) accumulator
-----------------------------
`chat()` accumulated per-id text via `buffers[id] = buffers.get(id,"") + delta`,
which is O(n) per concat → O(n²) total over a streamed response. At
~2 KB cumulative text this becomes user-visible; at 50 KB / 5000 chunks
it costs roughly 100-300 ms of pure copying. Switched to
`dict[str, list[str]]` + `"".join()` once at return.
Cleanup
-------
- Extract `_serialize_tool_calls`, `_ai_text_event`, `_ai_tool_calls_event`,
and `_tool_message_event` static helpers. The messages-mode and
values-mode branches previously repeated four inline dict literals each;
they now call the same builders.
- `StreamEvent.type` is now typed as `Literal["values", "messages-tuple",
"custom", "end"]` via a `StreamEventType` alias. Makes the closed set
explicit and catches typos at type-check time.
- Direct attribute access on `AIMessage`/`AIMessageChunk`: `.usage_metadata`,
`.tool_calls`, `.id` all have default values on the base class, so the
`getattr(..., None)` fallbacks were dead code. Removed from the hot
path.
- `_account_usage` parameter type loosened to `Any` so that LangChain's
`UsageMetadata` TypedDict is accepted under strict type checking.
- Trimmed narrating comments on `seen_ids` / `streamed_ids` / the
values-synthesis skip block; kept the non-obvious ones that document
the cross-mode dedup invariant.
Net diff: -15 lines. All 132 unit tests + harness boundary test still
pass; ruff check and ruff format pass.
* docs(backend): add STREAMING.md design note (#1969)
Dedicated design document for the token-level streaming architecture,
prompted by the bug investigation in #1969.
Contents:
- Why two parallel streaming paths exist (Gateway HTTP/async vs
DeerFlowClient sync/in-process) and why they cannot be merged.
- LangGraph's three-layer mode naming (Graph "messages" vs Platform
SDK "messages-tuple" vs HTTP SSE) and why a shared string constant
would be harmful.
- Gateway path: run_agent + StreamBridge + sse_consumer with a
sequence diagram.
- DeerFlowClient path: sync generator + direct yield, delta semantics,
chat() accumulator.
- Why the three id sets (seen_ids / streamed_ids / counted_usage_ids)
each carry an independent invariant and cannot be collapsed.
- End-to-end sequence for a real conversation turn.
- Lessons from #1969: why mock-based tests missed the bug, why
BPE subword boundaries in live output are the strongest
correctness signal, and the regression test that locks it in.
- Source code location index.
Also:
- Link from backend/CLAUDE.md Embedded Client section.
- Link from backend/docs/README.md under Feature Documentation.
* test(backend): add refactor regression guards for stream() (#1969)
Three new tests in TestStream that lock the contract introduced by
PR #1974 so any future refactor (sync->async migration, sharing a
core with Gateway's run_agent, dedup strategy change) cannot
silently change behavior.
- test_dedup_requires_messages_before_values_invariant: canary that
documents the order-dependence of cross-mode dedup. streamed_ids
is populated only by the messages branch, so values-before-messages
for the same id produces duplicate AI text events. Real LangGraph
never inverts this order, but a refactor that does (or that makes
dedup idempotent) must update this test deliberately.
- test_messages_mode_golden_event_sequence: locks the *exact* event
sequence (4 events: 2 messages-tuple deltas, 1 values snapshot, 1
end) for a canonical streaming turn. List equality gives a clear
diff on any drift in order, type, or payload shape.
- test_chat_accumulates_in_linear_time: perf canary for the O(n^2)
fix in commit 1f11ba10. 10,000 single-char chunks must accumulate
in under 1s; the threshold is wide enough to pass on slow CI but
tight enough to fail if buffer = buffer + delta is restored.
All three tests pass alongside the existing 12 TestStream tests
(15/15). ruff check + ruff format clean.
* docs(backend): clarify stream() docstring on JSON serialization (#1969)
Replace the misleading "raw LangChain objects (AIMessage,
usage_metadata as dataclasses), not dicts" claim in the
"Why not reuse Gateway's run_agent?" section. The implementation
already yields plain Python dicts (StreamEvent.data is dict, and
usage_metadata is a TypedDict), so the original wording suggested
a richer return type than the API actually delivers.
The corrected wording focuses on what is actually true and
relevant: this client skips the JSON/SSE serialization layer that
Gateway adds for HTTP wire transmission, and yields stream event
payloads directly as Python data structures.
Addresses Copilot review feedback on PR #1974.
* test(backend): document none-id messages dedup limitation (#1969)
Add test_none_id_chunks_produce_duplicates_known_limitation to
TestStream that explicitly documents and asserts the current
behavior when an LLM provider emits AIMessageChunk with id=None
(vLLM, certain custom backends).
The cross-mode dedup machinery cannot record a None id in
streamed_ids (guarded by ``if msg_id:``), so the values snapshot's
reassembled AIMessage with a real id falls through and synthesizes
a duplicate AI text event. The test asserts len == 2 and locks
this as a known limitation rather than silently letting future
contributors hit it without context.
Why this is documented rather than fixed:
* Falling back to ``metadata.get("id")`` does not help — LangGraph's
messages-mode metadata never carries the message id.
* Synthesizing ``f"_synth_{id(msg_chunk)}"`` only helps if the
values snapshot uses the same fallback, which it does not.
* A real fix requires provider cooperation (always emit chunk ids)
or content-based dedup (false-positive risk), neither of which
belongs in this PR.
If a real fix lands, replace this test with a positive assertion
that dedup works for None-id chunks.
Addresses Copilot review feedback on PR #1974 (client.py:515).
* fix(frontend): UI polish - fix CSS typo, dark mode border, and hardcoded colors (#1942)
- Fix `font-norma` typo to `font-normal` in message-list subtask count
- Fix dark mode `--border` using reddish hue (22.216) instead of neutral
- Replace hardcoded `rgb(184,184,192)` in hero with `text-muted-foreground`
- Replace hardcoded `bg-[#a3a1a1]` in streaming indicator with `bg-muted-foreground`
- Add missing `font-sans` to welcome description `<pre>` for consistency
- Make case-study-section padding responsive (`px-4 md:px-20`)
Closes#1940
* docs: clarify deployment sizing guidance (#1963)
* fix(frontend): prevent stale 'new' thread ID from triggering 422 history requests (#1960)
After history.replaceState updates the URL from /chats/new to
/chats/{UUID}, Next.js useParams does not update because replaceState
bypasses the router. The useEffect in useThreadChat would then set
threadIdFromPath ('new') as the threadId, causing the LangGraph SDK
to call POST /threads/new/history which returns HTTP 422 (Invalid
thread ID: must be a UUID).
This fix adds a guard to skip the threadId update when
threadIdFromPath is the literal string 'new', preserving the
already-correct UUID that was set when the thread was created.
* fix(frontend): avoid using route new as thread id (#1967)
Co-authored-by: luoxiao6645 <luoxiao6645@gmail.com>
* Fix(subagent): Event loop conflict in SubagentExecutor.execute() (#1965)
* Fix event loop conflict in SubagentExecutor.execute()
When SubagentExecutor.execute() is called from within an already-running
event loop (e.g., when the parent agent uses async/await), calling
asyncio.run() creates a new event loop that conflicts with asyncio
primitives (like httpx.AsyncClient) that were created in and bound to
the parent loop.
This fix detects if we're already in a running event loop, and if so,
runs the subagent in a separate thread with its own isolated event loop
to avoid conflicts.
Fixes: sub-task cards not appearing in Ultra mode when using async parent agents
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(subagent): harden isolated event loop execution
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
* refactor(backend): remove dead getattr in _tool_message_event
---------
Co-authored-by: greatmengqi <chenmengqi.0376@bytedance.com>
Co-authored-by: Xinmin Zeng <135568692+fancyboi999@users.noreply.github.com>
Co-authored-by: 13ernkastel <LennonCMJ@live.com>
Co-authored-by: siwuai <458372151@qq.com>
Co-authored-by: 肖 <168966994+luoxiao6645@users.noreply.github.com>
Co-authored-by: luoxiao6645 <luoxiao6645@gmail.com>
Co-authored-by: Saber <11769524+hawkli-1994@users.noreply.github.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
* test(skills): add trigger eval set for systematic-literature-review skill
20 eval queries (10 should-trigger, 10 should-not-trigger) for use with
skill-creator's run_eval.py. Includes real-world SLR queries contributed
by @VANDRANKI (issue #1862 author) and edge cases for routing
disambiguation with academic-paper-review.
* test(skills): add grader expectations for SLR skill evaluation
5 eval cases with 39 expectations covering:
- Standard SLR flow (APA/BibTeX/IEEE format selection)
- Keyword extraction and search behavior
- Subagent dispatch for metadata extraction
- Report structure (themes, convergences, gaps, per-paper annotations)
- Negative case: single-paper routing to academic-paper-review
- Edge case: implicit SLR without explicit keywords
* refactor(skills): shorten SLR description for better trigger rate
Reduce description from 833 to 344 chars. Key changes:
- Lead with "systematic literature review" as primary trigger phrase
- Strengthen single-paper exclusion: "Not for single-paper tasks"
- Remove verbose example patterns that didn't improve routing
Tested with run_eval.py (10 runs/query):
- False positive "best paper on RL": 67% → 20% (improved)
- True positive explicit SLR query: ~30% (unchanged)
Low recall is a routing-layer limitation, not a description issue —
see PR description for full analysis.
* Potential fix for pull request finding
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
The /api/langgraph/* endpoints proxy straight to the LangGraph server,
so clients inherit LangGraph's native recursion_limit default of 25
instead of the 100 that build_run_config sets for the Gateway and IM
channel paths. 25 is too low for plan-mode or subagent runs and
reliably triggers GraphRecursionError on the lead agent's final
synthesis step after subagents return.
Set recursion_limit: 100 in the Create Run example and the cURL
snippet, and add a short note explaining the discrepancy so users
following the docs don't hit the 25-step ceiling as a surprise.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat(skills): add systematic-literature-review skill for multi-paper SLR workflows
Adds a new skill that produces a structured systematic literature review (SLR)
across multiple academic papers on a topic. Addresses #1862 with a pure skill
approach: no new tools, no architectural changes, no new dependencies.
Skill layout:
- SKILL.md — 4+1 phase workflow (plan, search, extract, synthesize, present)
- scripts/arxiv_search.py — arXiv API client, stdlib only, with a
requests->urllib fallback shim modeled after github-deep-research's
github_api.py
- templates/{apa,ieee,bibtex}.md — citation format templates selected
dynamically in Phase 4, mirroring podcast-generation's templates/ pattern
Design notes:
- Multi-paper synthesis uses the existing `task` tool to dispatch extraction
subagents in parallel. SKILL.md's Phase 3 includes a fixed decision table
for batch splitting to respect the runtime's MAX_CONCURRENT_SUBAGENTS = 3
cap, and explicitly tells the agent to strip the "Task Succeeded. Result: "
prefix before parsing subagent JSON output.
- arXiv only, by design. Semantic Scholar and PubMed adapters would push the
scope toward a standalone MCP server (see #933) and are intentionally out
of scope for this skill.
- Coexists with the existing `academic-paper-review` skill: this skill does
breadth-first synthesis across many papers, academic-paper-review does
single-paper peer review. The two are routed via distinct triggers and
can compose (SLR on many + deep review on 1-2 important ones).
- Hard upper bound of 50 papers, tied to the Phase 3 concurrency strategy.
Larger surveys degrade in synthesis quality and are better split by
sub-topic.
BibTeX template explicitly uses @misc for arXiv preprints (not @article),
which is the most common mistake when generating BibTeX for arXiv papers.
arxiv_search.py was smoke-tested end-to-end against the live arXiv API with
two query shapes (relevance sort, submittedDate sort with category filter);
all returned JSON fields parse correctly (id normalization, Atom namespace
handling, URL encoding for multi-word queries).
* fix(skills): prevent LLM from saving intermediate search results to file
Adds an explicit "do not save" instruction at the end of Phase 2.
Observed during Test 1 with DeepSeek: the model saved search results
to a markdown file before proceeding to Phase 3, wasting 2-3 tool call
rounds and increasing the risk of hitting the graph recursion limit.
The search JSON should stay in context for Phase 3, not be persisted.
* fix(skills): use relevance+start-date instead of submittedDate sorting
Test 2 revealed that arXiv's submittedDate sorting returns the most
recently submitted papers in the category regardless of query relevance.
Searching "diffusion models" with sortBy=submittedDate in cs.CV returned
papers on spatial memory, Navier-Stokes, and photon-counting CT — none
about diffusion models. The LLM then retried with 4 different queries,
wasting tool calls and approaching the recursion limit.
Fix: always sort by relevance; when the user wants "recent" papers,
combine relevance sorting with --start-date to constrain the time window.
Also add an explicit "run the search exactly once" instruction to prevent
the retry loop.
* fix(skills): wrap multi-word arXiv queries in double quotes for phrase matching
Without quotes, `all:diffusion model` is parsed by arXiv's Lucene as
`all:diffusion OR model`, pulling in unrelated papers from physics
(thermal diffusion) and other fields. Wrapping in double quotes forces
phrase matching: `all:"diffusion model"`.
Also fixes date filtering: the previous bug caused 2011 papers to appear
in results despite --start-date 2024-04-09, because the unquoted query
words were OR'd with the date constraint.
Verified: "diffusion models" --category cs.CV --start-date 2024-04-09
now returns only relevant diffusion model papers published after April
2024.
* fix(skills): add query phrasing guide and enforce subagent delegation
Two fixes from Test 2 observations with DeepSeek:
1. Query phrasing: add a table showing good vs bad query examples.
The script wraps multi-word queries in double quotes for phrase
matching, so long queries like "diffusion models in computer vision"
return 0 results. Guide the LLM to use 2-3 core keywords + --category
instead.
2. Subagent enforcement: DeepSeek was extracting metadata inline via
python -c scripts instead of using the task tool. Strengthen Phase 3
to explicitly name the task tool, say "do not extract metadata
yourself", and explain why (token budget, isolation). This is more
direct than the previous natural-language-only approach while still
providing the reasoning behind the constraint.
* fix(skills): strengthen search keyword guidance and subagent enforcement
Address two issues found during end-to-end testing with DeepSeek:
1. Search retry: LLM passed full topic descriptions as queries (e.g.
"diffusion models in computer vision"), which returned 0 results due
to exact phrase matching and triggered retries. Added explicit
instruction to extract 2-3 core keywords before searching.
2. Subagent bypass: LLM used python -c to extract metadata instead of
dispatching via task tool. Added explicit prohibition list (python -c,
bash scripts, inline extraction) with ❌ markers for clarity.
* fix(skills): address Copilot review feedback on SLR skill
- Fix legacy arXiv ID parsing: preserve archive prefix for pre-2007
papers (e.g. hep-th/9901001 instead of just 9901001)
- Fix phase count: "four phases" -> "five phases"
- Add subagent_enabled prerequisite note to SKILL.md Notes section
- Remove PR-specific references ("PR 1") from ieee.md and bibtex.md
templates, replace with workflow-scoped wording
- Fix script header: "stdlib only" -> "no additional dependencies
required", fix relative path to github_api.py reference
- Remove reference to non-existent docs/enhancement/ path in header
* Apply suggestions from code review
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
---------
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Add langchain-ollama as an optional dependency and provide ChatOllama
config examples, enabling proper thinking/reasoning content preservation
for local Ollama models.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(frontend): replace invalid "context" select field with "metadata" in threads.search
The LangGraph API server does not support "context" as a select field for
threads/search, causing a 422 Unprocessable Entity error introduced by
commit 60e0abf (#1771).
- Replace "context" with "metadata" in the default select list
- Persist agent_name into thread metadata on creation so search results
carry the agent identity
- Update pathOfThread() to fall back to metadata.agent_name when
context is unavailable from search results
- Add regression tests for metadata-based agent routing
Fixes#2037
Made-with: Cursor
* fix: apply Copilot suggestions
* style: fix the lint error
* feat(config): add when_thinking_disabled support for model configs
Allow users to explicitly configure what parameters are sent to the
model when thinking is disabled, via a new `when_thinking_disabled`
field in model config. This mirrors the existing `when_thinking_enabled`
pattern and takes full precedence over the hardcoded disable behavior
when set. Backwards compatible — existing configs work unchanged.
Closes#1675
* fix(config): address copilot review — gate when_thinking_disabled independently
- Switch truthiness check to `is not None` so empty dict overrides work
- Restructure disable path so when_thinking_disabled is gated independently
of has_thinking_settings, allowing it to work without when_thinking_enabled
- Update test to reflect new behavior
* feat: implement full checkpoint rollback on user cancellation
- Capture pre-run checkpoint snapshot including checkpoint state, metadata, and pending_writes
- Add _rollback_to_pre_run_checkpoint() function to restore thread state
- Implement _call_checkpointer_method() helper to support both async and sync checkpointer methods
- Rollback now properly restores checkpoint, metadata, channel_versions, and pending_writes
- Remove obsolete TODO comment (Phase 2) as rollback is now complete
This resolves the TODO(Phase 2) comment and enables full thread state
restoration when a run is cancelled by the user.
* fix: address rollback review feedback
* fix: strengthen checkpoint rollback validation and error handling
- Validate restored_config structure and checkpoint_id before use
- Raise RuntimeError on malformed pending_writes instead of silent skip
- Normalize None checkpoint_ns to empty string instead of "None"
- Move delete_thread to only execute when pre_run_snapshot is None
- Add docstring noting non-atomic rollback as known limitation
This addresses review feedback on PR #1867 regarding data integrity
in the checkpoint rollback implementation.
* test: add comprehensive coverage for checkpoint rollback edge cases
- test_rollback_restores_snapshot_without_deleting_thread
- test_rollback_deletes_thread_when_no_snapshot_exists
- test_rollback_raises_when_restore_config_has_no_checkpoint_id
- test_rollback_normalizes_none_checkpoint_ns_to_root_namespace
- test_rollback_raises_on_malformed_pending_write_not_a_tuple
- test_rollback_raises_on_malformed_pending_write_non_string_channel
- test_rollback_propagates_aput_writes_failure
Covers all scenarios from PR #1867 review feedback.
* test: format rollback worker tests
* fix(sandbox): add startup reconciliation to prevent orphaned container leaks
Sandbox containers were never cleaned up when the managing process restarted,
because all lifecycle tracking lived in in-memory dictionaries. This adds
startup reconciliation that enumerates running containers via `docker ps` and
either destroys orphans (age > idle_timeout) or adopts them into the warm pool.
Closes#1972
* fix(sandbox): address Copilot review — adopt-all strategy, improved error handling
- Reconciliation now adopts all containers into warm pool unconditionally,
letting the idle checker decide cleanup. Avoids destroying containers
that another concurrent process may still be using.
- list_running() logs stderr on docker ps failure and catches
FileNotFoundError/OSError.
- Signal handler test restores SIGTERM/SIGINT in addition to SIGHUP.
- E2E test docstring corrected to match actual coverage scope.
* fix(sandbox): address maintainer review — batch inspect, lock tightening, import hygiene
- _reconcile_orphans(): merge check-and-insert into a single lock acquisition
per container to eliminate the TOCTOU window.
- list_running(): batch the per-container docker inspect into a single call.
Total subprocess calls drop from 2N+1 to 2 (one ps + one batch inspect).
Parse port and created_at from the inspect JSON payload.
- Extract _parse_docker_timestamp() and _extract_host_port() as module-level
pure helpers and test them directly.
- Move datetime/json imports to module top level.
- _make_provider_for_reconciliation(): document the __new__ bypass and the
lockstep coupling to AioSandboxProvider.__init__.
- Add assertion that list_running() makes exactly ONE inspect call.
* Fix HTML artifact preview rendering
* Add after screenshot for HTML preview fix
* Add before screenshot for HTML preview fix
* Update before screenshot for HTML preview fix
* Update after screenshot for HTML preview fix
* Update before screenshot to Tsinghua homepage repro
* Update after screenshot to Tsinghua homepage preview
* Address PR review on HTML artifact preview
* Harden HTML artifact preview isolation
Streamdown's streaming safeguard appends closing markers (e.g. `*`) to
text with unmatched markdown syntax. This causes user messages containing
literal `*` (such as `99 * 87`) to display with a spurious trailing
asterisk. Human messages are always complete, so the incomplete-markdown
pre-processing is unnecessary.
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* fix(docker): nginx fails to start on hosts without IPv6
- Detect IPv6 support at runtime and remove `listen [::]` directive
when unavailable, preventing nginx startup failure on non-IPv6 hosts
- Use `exec` to replace shell with nginx as PID 1 for proper signal
handling (graceful shutdown on SIGTERM)
- Reformat command from YAML folded scalar to block scalar (no
functional change)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix(docker): harden nginx startup script (Copilot review feedback)
Add `set -e` so envsubst failures exit immediately instead of starting
nginx with an incomplete config. Narrow the sed pattern to match only
the `listen [::]:2026;` directive to avoid accidentally removing future
lines containing [::].
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
When a model config includes `reasoning_effort` as an extra YAML field
(ModelConfig uses `extra="allow"`), and the thinking-disabled code path
also injects `reasoning_effort="minimal"` into kwargs, the previous
`model_class(**kwargs, **model_settings_from_config)` call raises:
TypeError: got multiple values for keyword argument 'reasoning_effort'
Fix by merging the two dicts before instantiation, giving runtime kwargs
precedence over config values: `{**model_settings_from_config, **kwargs}`.
Fixes#1977
Co-authored-by: octo-patch <octo-patch@github.com>
