mirror of
https://github.com/bytedance/deer-flow.git
synced 2026-09-25 14:06:18 +00:00
9 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
Wenchao An
|
ca23703ef0
|
fix(subagents): preserve actionable acceptance gaps after compaction (#5287)
* fix: preserve actionable subagent acceptance gaps Distinguish completed execution from acceptance in delegation guidance. Retain bounded unmet and unverified criteria after compaction and guide the lead to address remaining work within its budget. * docs: keep acceptance guidance within instruction budget |
||
|
Zeren Wang
|
a06a6fed7e
|
feat(harness): deterministic acceptance checklist for subagent delegations (RFC #4651, layer 2) (#5109)
* feat(harness): deterministic acceptance checklist for subagent delegations (RFC #4651, layer 2) PR4 of RFC #4651: check lead-supplied acceptance_criteria in code when a subagent completes, so objectively checkable requirements can never be silently passed by a self-report. - subagents/acceptance_checks.py: deterministic leaf families — file:<path> exists|non-empty and file_written:<path> read through read_current_file_content scoped to the shared thread workspace; the read uses the sandbox-native virtual path form (the local read validator and provider mount tables resolve /mnt/user-data/... paths, not host paths); the scope decision canonicalizes with realpath on the local sandbox so workspace symlinks cannot escape into uploads; a remote provider's "Error: ..." return string is normalized to a failed check (provider-typed via is_local_sandbox); a UnicodeDecodeError marks a binary deliverable as existing and non-empty; out-of-scope paths degrade to UNVERIFIED. tests_passed:<command> anchors to a matching recorded bash execution with status=success and a test-summary shape; matching is shell-structure aware with control-flow attribution (span must end at the last segment with provable execution), negating-option values are ineligible evidence and a target negated anywhere in the command degrades the match, extra flags must be selection-preserving, extra positionals widen only after a path-scoped criterion, truncated commands degrade via command_truncated, the summary shape is read only from output attributable to the matched segment (preceding segments provably silent by invocation form), and pass shapes require a nonzero passed count. Criterion text is neutralized with neutralize_untrusted_tags before storage/rendering. Anything else renders UNVERIFIED, never silently passed. - executor: accumulate bounded bash command/output evidence per streamed chunk (merged by tool_call_id, newest-capped) so subagent summarization compacting earlier messages cannot erase a recorded execution; the recorded status is the actual shell exit status parsed from the output's exit marker (signed codes included; the remote Command exited with code N form is accepted only as the whole trimmed output), falling back to deerflow_tool_meta only when no marker exists. - sandbox providers: e2b/opensandbox/tenki/boxlite append the LocalSandbox-style "Exit Code: N" marker on nonzero exit even with non-empty output; aio propagates the SDK's structured exit_code on both exec paths the same way; local timeouts append Exit Code: 124; and _truncate_bash_output always preserves a trailing exit marker (signed included) inside its budget, with a 32-char floor raising any smaller configured limit, so the actual shell outcome always survives in the output text. - task_tool: run the checklist offloaded (asyncio.to_thread) on the completed branch, failure-isolated; stamp the verdict into result metadata and render the per-criterion section into the model-visible result text. - status contract: additive subagent_acceptance_verdict transport with read-side structural validation. - delegation ledger: entry carries the verdict and renders a compact acceptance segment; gateway strips caller-forged verdicts from both ledger entries and message metadata, like the citation verdict. - blocking-IO anchor pins the offload (teeth proven red->green); leaf read errors catch only OSError/SandboxError so unexpected errors reach the task-tool-level isolation instead of being mislabeled. * fix(harness): close acceptance evidence gaps from review (RFC #4651 PR4) - negating options: overlap with a matched criterion target is now checked by path/nodeid prefix, not exact token equality — excluding a sub-path of the criterion's selection (pytest tests --deselect tests/unit/test_auth.py) degrades to UNVERIFIED instead of holds - output attribution: any redirection token in the matched final segment makes the recorded tail non-attributable (> / >> / 2> are word characters to the parser, so redirection was invisible to the matcher) - silent-source allowlist narrowed from any *activate suffix to the */bin/activate shape - status_contract docstring: restore the shared-fixture sentence and note subagent_acceptance_verdict is deliberately outside the fixture - executor: update_bash_executions publishes [] (stream carried no bash-family calls) instead of collapsing it into None, mirroring update_tool_receipts * fix(harness): close acceptance residual gaps from re-review (RFC #4651 PR4) - tests_passed: add error outcomes to the fail shapes — "4 passed, 1 error" and pytest's "ERROR <nodeid>" short summary no longer satisfy the pass shape when the exit status is swallowed (|| true) or absent; zero-error counts stay clean. - file leaves: bound the deliverable read — a "wc -c" shell size probe answers files above 50k bytes without loading ~2x their size, honoring the host-bash kill switch and falling back to the full read on any non-integer rendering, so verdicts never get less sound. - executor: record the exit marker text as status_marker on harvested bash evidence; the leaf detail now reports the marker actually seen instead of asserting a failure indistinguishable from the command's own trailing text. - extend the blocking-IO anchor to drive the probe branch inside the offload; teeth re-verified red->green. * fix(harness): close acceptance forgery and bound gaps from P2 re-review (RFC #4651 PR4) - file leaves: never read unbounded — size is established first (os.stat on the validated local host path, so the host-bash-disabled configuration needs no shell; a guarded wc -c on remote providers that renders missing/unreadable in its own words). Above the 50k cap the leaf answers from the size alone, at/below it the full read runs, and an unestablishable size degrades to UNVERIFIED instead of an unlimited fallback read. - output attribution: source/. prefixes are never provably silent — a crafted */bin/activate path shape says nothing about what the script prints, so sourced segments can no longer lend a passing summary. - executable identity: an explicitly path-spelled criterion now requires the same normalized executable path; the basename rule stays only for deliberately bare criterion commands. * fix(harness): run acceptance size probe outside subagent-controlled state (RFC #4651 PR4) - remote probe no longer runs in the sandbox's persistent shell: a fresh env -i /bin/sh with absolute-path stat/realpath (poisoned functions, aliases, PATH, exported functions, IFS, locale cannot steer it), plus a marker env routing AIO onto a fresh per-call bash.exec session. - metadata-only: stat never opens content, so a FIFO deliverable cannot block the parent for the provider's idle timeout; non-regular files (fifo/dir/symlink) degrade to UNVERIFIED. - containment canonicalized against the literal mount root: a final-component symlink or a swapped parent directory (root included) cannot redirect the check outside shared storage; unprovable layouts degrade to UNVERIFIED. * fix(harness): canonicalize probe containment against the canonical mount root (RFC #4651 PR4) Literal-root equality made every remote file leaf permanently UNVERIFIED on e2b and Tenki, which realize /mnt/user-data as a symlink to the home dir by default (e2b bootstrap 'sudo ln -sfn', Tenki best-effort symlink). Containment now compares the file's realpath against the mount root's realpath — exactly what the provider's own read path resolves, so probe and read-back stay consistent; final-component symlinks stay rejected by the non-dereferencing stat, and an intermediate dir-link escape under a sane root still lands ESCAPED. The inner script is a module constant and the suite now executes the composed probe for real against on-disk layouts (real dir, symlinked prefix, final symlink, fifo, missing, dir-link escape), which the canned-output stub could not see. * fix(harness): close bare-criterion negation and CDPATH summary channels (RFC #4651 PR4) - matching: a criterion with no positional selection target (bare pytest, make test) stands for the runner's default selection, so ANY negating option (--ignore/--deselect/...) makes the recorded run a different selection — unprovable. The overlap guard only sees consumed criterion tokens, which a bare criterion does not have; scoped criteria keep the unrelated-exclusion behavior. - attribution: cd is no longer blanket-silent — CDPATH makes cd print the resolved (subagent-chosen) destination and the pass shapes match as substrings, so one mkdir 'all tests passed' plus an export minted a pass for any quiet command. A cd argument or CDPATH= value (export or leading assignment) carrying any summary shape makes the segment non-silent; shape-free cd dir wrappers keep matching. - docs: _truncate_bash_output states the effective 32-char floor (the guarantee previously read as an unconditional max_chars bound). * fix(harness): close env-assignment and expansion channels in acceptance matching (RFC #4651 PR4) Self-audit in the shape of the last review rounds — channels the matcher classified as accounted-for that can change what runs, narrow the selection, or lend the summary text: - env assignments are no longer blanket-stripped: only an allowlist of inert display/CI knobs (CI, NO_COLOR, PY_COLORS, ...) may prefix a matched span, and a non-allowlisted assignment in any preceding segment (pure-assignment or export NAME=) is state pollution — PATH redirects the executable, LD_PRELOAD/PYTHONPATH/NODE_OPTIONS inject code, PYTEST_ADDOPTS/GOFLAGS/MAKEFILES inject selection-changing inputs, BASH_ENV runs arbitrary shell startup. All degrade to unprovable. - runtime expansions: any span token carrying /$( )/backticks, any negating-option value carrying an expansion or glob (unknown excluded set), and any extra executed token carrying glob metacharacters (crafted option-looking filenames narrow invisibly) are unprovable. Criterion-side globs stay self-consistent (literal match). - cd: an argument carrying a runtime expansion or glob is non-silent (unknown destination, unknown print); CDPATH= assignments are now handled as state pollution at the match layer, subsuming the value-shape special case. * fix(harness): persistent-shell evidence, exact env sets, option-arity scoping (RFC #4651 PR4) - tests_passed: on a persistent-shell provider (new Sandbox.persistent_shell_sessions capability, set by AioSandbox) every leaf degrades to UNVERIFIED — any earlier call in the shared session could have mutated the state the clean-looking run executed in, and only a fresh controlled session (RFC section 6 verifier) can prove otherwise. The flag is read from the provider registry without acquiring a sandbox. - env assignments: the allowlist is gone — no variable is provably inert across repositories (CI/DEBUG are routinely read by tests). The span's assignment prefix must equal the criterion's exactly (values included, order-insensitive); any assignment or export NAME= in a preceding segment is state pollution. - scoping: positional targets are now read by option arity, so a path embedded in an option (--basetemp=/tmp/p, --junitxml=/tmp/r.xml) never counts as a selection target and an extra positional after such a criterion narrows the default selection it denotes. * fix(harness): stamp shell provenance at harvest, close export/unset and arity gaps (RFC #4651 PR4) * fix(harness): split physical newlines as shell separators in acceptance matching (RFC #4651 PR4) * fix(harness): scope cd wrappers to thread data roots, pin accepted boundaries (RFC #4651 PR4) * fix(harness): preserve criterion connectors, prove file_written readable, fail-closed shell capability (RFC #4651 PR4) * fix(harness): compare only the connector prefix, tolerate trailing criterion semicolons (RFC #4651 PR4) * fix(harness): preserve continuation-line operators, keep ./-spelled executable identity (RFC #4651 PR4) * fix(harness): render criteria single-line so a multiline criterion cannot inject a forged checklist line (RFC #4651 PR4) * fix(harness): reject parent-traversal executable tokens in acceptance matching (RFC #4651 PR4) * fix(harness): reject parent-traversal negated values in acceptance matching (RFC #4651 PR4) |
||
|
Zeren Wang
|
3b592c2053
|
feat(harness): subagent receipt citation verification (#5076)
* feat(harness): subagent receipt citation verification - add receipt citation verification core - harvest subagent tool receipts at terminal status - transport subagent receipts and citation verdict via status contract - verify subagent report citations at task write-back - render citation verdicts in the delegation ledger * fix(gateway): strip forged receipt verdicts from the delegations channel normalize_input() and the checkpoint-state mutation sanitizer only stripped server-owned metadata from message-shaped values, so an external caller could submit a delegation entry carrying a forged receipt_verdict that render_delegation_ledger would present as runtime-owned execution evidence. Strip receipt_verdict from caller-supplied delegation entries on both the run-creation and thread-state mutation paths, with regression coverage for each. * fix(harness): close silent-pass gaps in the zero-citation heuristic The action-claim detector missed the most common completion verbs (fixed/added/tested/changed/...) and had no CJK coverage at all, so reports like 'I fixed the bug and added tests.' or '我已经创建了文件并运行了测试。' were treated as claim-free: citation_resolved=True with no ledger warning. Broaden the verb lists and add a language-independent safety net: when the run harvested receipts and a nontrivial (>=240 char) report cites none of them, flag it UNVERIFIED. Short claim-free confirmations remain a vacuous pass. |
||
|
DanielWalnut
|
4e209827f3
|
feat(agent): Add subagent total delegation cap (#4115)
* fix subagent total delegation cap * fix embedded subagent run cap context * fix subagent cap config consistency * fix resumed subagent run cap boundary * fix legacy resume subagent boundary * address subagent cap review feedback --------- Co-authored-by: Willem Jiang <willem.jiang@gmail.com> |
||
|
hataa
|
c9fb9768d4
|
fix(subagents): unify guardrail caps on additive stop_reason + add token_budget (#3875 Phase 2) (#3980)
Phase 2 of #3875. Two guardrail axes can end a subagent run early — the turn budget (GraphRecursionError) and the token budget (TokenBudgetMiddleware) — and both now surface *why* through one additive `subagent_stop_reason` field instead of a status enum. This completes and course-corrects Phase 1 (#3949), which shipped the turn-budget cap as a `max_turns_reached` status enum. The agreed Phase 2 design replaces that enum with an optional `stop_reason` field (token_capped | turn_capped | loop_capped): a new enum value would break v1 consumers, while an additive field is ignored by older frontends and ledger readers. `max_turns_reached` and SubagentStatus.MAX_TURNS_REACHED are removed. - subagents.token_budget config (default enabled, 2,000,000 tokens, warn 0.7) with per-agent override; TokenBudgetMiddleware is now attached in build_subagent_runtime_middlewares so the cost-ceiling backstop engages for every subagent. The hard-stop does not raise — it strips tool_calls and lets the run finish with a final answer, recording the cap on a per-run consume_stop_reason() accessor. - executor.py: on normal completion it reads consume_stop_reason() and stamps completed + token_capped when the budget fired; on GraphRecursionError it recovers the last AIMessage partial (completed + turn_capped) or, if nothing usable survived, failed + turn_capped. SubagentResult gains stop_reason. - status_contract.py / contracts/subagent_status_contract.json (v2) / frontend subtask-result.ts: additive subagent_stop_reason field, pinned by test_status_values_match_contract / test_stop_reason_values_match_contract. - task_tool.py + delegation_ledger.py: drop the max_turns_reached paths; the ledger captures stop_reason and renders model-facing "capped" guidance so the lead reuses a capped completion knowingly. The 2,000,000-token default is deliberately loose (tighten to taste) — it would have roughly halved the reported 4.4M burn while leaving legitimate deep-research runs (max_turns=150) room. Subagent summarization is a follow-up. |
||
|
hataa
|
0664ea2243
|
fix(subagents): surface turn-budget cap as MAX_TURNS_REACHED with partial result (#3875 Phase 2) (#3949)
* fix(subagents): surface turn-budget cap as MAX_TURNS_REACHED with partial result (#3875) Phase 2 of #3875. When a subagent exhausts its turn budget (recursion_limit == max_turns), LangGraph raises GraphRecursionError from agent.astream. The generic except Exception in _aexecute misclassified it as FAILED and discarded the partial work already streamed into final_state, so the lead could not tell 'broken subagent' from 'out of budget' and got an empty failure. Catch GraphRecursionError specifically (before the generic handler) and set a distinct SubagentStatus.MAX_TURNS_REACHED terminal status, recovering the partial result from the last streamed chunk via a shared _extract_final_result helper (refactored out of the normal-completion path so both paths render content identically). Extend the cross-language status contract so the new value travels on additional_kwargs.subagent_status: a capped run is result-bearing, so make_subagent_additional_kwargs / read_subagent_result_metadata carry subagent_result_brief + subagent_result_sha256 (the recovered work, like completed) AND the cap notice on subagent_error -- the one status that carries both. task_tool.py returns it via the shared _task_result_command; the delegation ledger prefers the partial result_brief and renders model-facing guidance (reuse / retry tighter / raise max_turns). Frontend collapses max_turns_reached to the failed pill with the cap notice on error. No agent-loop, runner, or persistence behavior touched; default max_turns is unchanged. * refactor(subagents): consolidate content-stringify onto shared helper Address review feedback on #3949 (willem-bd, copilot-pull-request-reviewer): - executor.py: drop the private `_stringify_message_content` — a third near-duplicate of `utils/messages.py::message_content_to_text`. `_extract_final_result` now delegates to that canonical helper; the "No response generated" sentinel is pushed down to the consumer (the shared helper returns "" for no-text, matching every other call site). - task_tool.py: align the live `task_failed` event's error string with the canonical "Reached max_turns=N" used by the logger, the structured `error=`, and the executor (was "Reached max turns (N)"). Behavior for real AIMessage content is unchanged; only atypical edge inputs (consecutive bare-string list items; empty content) now match the canonical helper that every other call site already uses. `extract_response_text` is intentionally left as-is: it filters by OpenAI content-block `type`, a different shape with many callers and its own tests. Co-Authored-By: Claude <noreply@anthropic.com> --------- Co-authored-by: Claude <noreply@anthropic.com> |
||
|
AochenShen99
|
66b9e7f212
|
feat: emit structured runtime metadata (follow-up#3887) (#3906)
* feat: emit structured runtime metadata * fix: avoid subagent import cycle in replay gateway * fix: preserve legacy subtask result parsing * refactor: tighten runtime metadata contracts * fix(middleware): keep recovery hint on task exception wrapper content The structured-metadata stamp overwrote the wrapper text with the bare task-failure message, dropping the model-facing 'Continue with available context, or choose an alternative tool.' guidance that every other tool exception keeps. Append the shared hint after the formatted message. * fix(subagents): require lowercase hex for result_sha256 reader Length-only validation accepted any 64-char string; a faulty serializer or relaying wrapper could store a non-digest value in the delegation ledger. Enforce the producer's hexdigest shape with a fullmatch. --------- Co-authored-by: Willem Jiang <willem.jiang@gmail.com> |
||
|
AochenShen99
|
442248dd06
|
feat: preserve durable context across summarization (#3887)
* feat: preserve durable context across summarization * fix: harden durable context review gaps * style: format delegation ledger live test * chore: remove stale delegation ledger prefix * fix: address durable context review feedback |
||
|
Nan Gao
|
af0d14d296
|
feat(subagents): system-maintained delegation ledger to stop redundant re-delegation (#3877)
* feat(subagents): add delegations ledger field + reducer to ThreadState * feat(subagents): pure helpers to derive + format the delegation ledger * feat(subagents): DelegationLedgerMiddleware records + injects the ledger * feat(subagents): register DelegationLedgerMiddleware for lead when subagents enabled + docs * add runtime log * chore(subagents): make delegation-ledger injection log production-ready * test(subagents): make delegation-ledger registration tests config-free; refresh ultra replay golden for delegations channel * refactor(subagents): derive TERMINAL_STATUSES from SUBAGENT_STATUS_VALUES + pin it Make thread_state's TERMINAL_STATUSES a frozenset over the status contract's SUBAGENT_STATUS_VALUES instead of a hardcoded literal, so the terminal-status set can never drift from the contract. Add a pinning test asserting the derivation and that the non-terminal "in_progress" stays excluded. Addresses PR #3877 review. --------- Co-authored-by: Willem Jiang <willem.jiang@gmail.com> |