fix(lobster): keep ordinary run/resume on default flow fields by LiLan0125 · Pull Request #102036 · openclaw/openclaw

LiLan0125 · 2026-07-08T04:56:14Z

Closes #102011

What Problem This Solves

Ordinary @openclaw/lobster run and resume calls could be routed into managed TaskFlow parsing when the tool payload contained schema-injected defaults such as flowStateJson: "{}" and flowExpectedRevision: 0.

Why This Change Was Made

The Lobster parser now distinguishes injected empty defaults from intentional managed fields:

blank flow state is treated as absent;
an empty object does not select managed mode by itself;
explicit managed runs still preserve stateJson: {};
revision 0 alone does not select managed mode;
explicit cross-mode fields are rejected before ordinary fallback.

This stays inside the Lobster plugin parser. It does not change core TaskFlow, protocol, configuration, migration, storage, or dependency contracts.

User Impact

Ordinary Lobster approval workflows can run and resume without being asked for TaskFlow-only fields the caller did not intend to provide. Managed TaskFlow callers retain existing validation and can explicitly start with an empty state object.

Evidence

Validated exact head c12bf2f48570da79968487e1dcc10fb4b9ac2f35.

Sanitized AWS Crabbox lease cbx_c92971815ef7, run run_f48d59554de8, exit 0; lease released after proof.
pnpm test extensions/lobster/src/lobster-tool.test.ts extensions/lobster/src/lobster-runner.test.ts: 2 files, 52 tests passed.
Real createLobsterTool(createEmbeddedLobsterRunner()) ordinary run -> approval -> resume with injected {} and revision 0: passed without a TaskFlow runtime.
Real embedded managed run retained an own stateJson property equal to {}.
Full remote pnpm check:changed: passed, including extension production/test typechecks, lint, database/storage guards, and import-cycle checks.
Hosted CI run 28920766065: 63 successful checks, no failures or pending checks.
Fresh autoreview after the maintainer strictness repair: clean, no actionable findings.
git diff --check: passed.

No documentation or changelog update is required; the public call shapes are unchanged and release generation owns CHANGELOG.md.

clawsweeper · 2026-07-08T04:58:41Z

Codex review: needs real behavior proof before merge. Reviewed July 8, 2026, 1:49 AM ET / 05:49 UTC.

Summary
The branch updates the Lobster tool parser and tests so blank or empty flowStateJson and flowExpectedRevision: 0 do not force ordinary run/resume calls into managed TaskFlow mode while preserving explicit empty managed run state.

PR surface: Source +16, Tests +125. Total +141 across 2 files.

Reproducibility: yes. source-level. Current main and v2026.6.11 route default flow fields through the managed-parser decision path before ordinary runner fallback; I did not run a live Gateway reproduction in this read-only review.

Review metrics: 1 noteworthy metric.

Schema-default routing: 2 default flow values handled. The PR changes how empty flowStateJson and flowExpectedRevision: 0 select ordinary versus managed Lobster execution.

Stored data model
Persistent data-model change detected: serialized state: extensions/lobster/src/lobster-tool.ts. Confirm migration or upgrade compatibility proof before merge.

Root-cause cluster
Relationship: fixed_by_candidate
Canonical: #102011
Summary: This PR is one candidate fix for the canonical Lobster schema-default flow-field misrouting issue; another open candidate has stronger proof and one weaker sibling was already closed unmerged.

Members:

canonical: @openclaw/lobster tool run/resume misroutes schema default flowExpectedRevision into TaskFlow mode #102011 - The PR body uses closing syntax for the same ordinary Lobster run/resume default flow-field misrouting report.
fixed_by_candidate: fix(lobster): don't route schema-default flowExpectedRevision into TaskFlow (#102011) #102022 - This open sibling PR targets the same canonical Lobster parser bug and currently has proof-sufficient, ready-for-maintainer-look status.
superseded: fix(lobster): treat schema-default flowExpectedRevision=0 as absent for ordinary run/resume #102023 - This closed unmerged sibling targeted the same issue but was already removed from the candidate set after review.
same_root_cause: Lobster: add managed TaskFlow mode #61555 - This merged PR added the managed TaskFlow parser path whose field-detection logic is implicated by the report.
partial_overlap: Lobster: harden embedded runtime integration #61566 - This merged PR hardened the same Lobster embedded runtime/parser surface but did not address injected default field intent.

Proposal only: this assessment does not dispatch repair, suppress jobs, mutate sibling items, close, or merge anything.

Merge readiness
Overall: 🦪 silver shellfish
Proof: 🦪 silver shellfish
Patch quality: 🐚 platinum hermit
Result: blocked until real behavior proof from a real setup is added.

Overall follows the weaker of proof and patch quality, so missing proof can cap an otherwise strong patch.

Rank-up moves:

[P1] Add redacted real installed-plugin, Gateway, or real embedded @clawdbot/lobster runtime proof for ordinary run then resume with default flow fields.
[P1] Have a maintainer choose the canonical landing candidate between this PR and fix(lobster): don't route schema-default flowExpectedRevision into TaskFlow (#102011) #102022 before merge.

Proof guidance:

[P1] Needs real behavior proof before merge: The terminal proof exercises patched createLobsterTool with fake runner and TaskFlow dependencies; add redacted installed-plugin, Gateway, or real embedded-runtime output, then update the PR body to trigger a fresh ClawSweeper review or ask a maintainer to comment @clawsweeper re-review.

Risk before merge

[P1] The contributor proof still uses fake runner and TaskFlow dependencies, so it does not prove a real installed @openclaw/lobster plugin, embedded runtime, or Gateway approval workflow after the fix.
[P1] fix(lobster): don't route schema-default flowExpectedRevision into TaskFlow (#102011) #102022 remains open for the same canonical bug with stronger real-runtime proof, so landing both would duplicate the same parser fix.
[P1] The live PR is mergeable but behind its recorded base and some checks were still in progress during review, so exact-head CI should refresh before any landing decision.

Maintainer options:

Decide the mitigation before merge
Land one canonical Lobster parser fix with real installed-plugin or Gateway proof that ordinary default-field run/resume stays on the runner path while managed TaskFlow mode remains intact, then close the duplicate candidates and linked issue after merge.
Pause or close
Do not merge this PR until maintainers decide whether the risk is worth taking.

Next step before merge

[P1] Manual review is needed because this fresh contributor PR competes with a proof-sufficient sibling and still needs contributor real-behavior proof; there is no narrow code repair for automation to apply here.

Maintainer decision needed

Question: Should maintainers use this PR or fix(lobster): don't route schema-default flowExpectedRevision into TaskFlow (#102011) #102022 as the canonical Lobster parser fix for @openclaw/lobster tool run/resume misroutes schema default flowExpectedRevision into TaskFlow mode #102011?
Rationale: Both PRs target the same shipped Lobster parser bug; this branch is focused and assigned but still has mock-only proof, while the sibling currently has stronger real-runtime proof and a ready-for-maintainer-look status.
Likely owner: vincentkoc — The live PR is assigned to this account and current parser blame plus recent Lobster/check history make them the clearest routing owner for the duplicate-candidate choice.
Options:
- Promote the proof-positive sibling (recommended): Use fix(lobster): don't route schema-default flowExpectedRevision into TaskFlow (#102011) #102022 as the canonical landing path if its latest head still clears maintainer review, then close this duplicate after that fix merges.
- Keep this PR as backup pending proof: Keep this PR open but do not merge it until the contributor adds redacted installed-plugin or Gateway proof and a maintainer compares it with the sibling candidate.
- Close duplicates after canonical merge: Wait until one candidate lands, then close the remaining open duplicate PRs and the linked issue with the landed commit as proof.

Security
Cleared: The diff is limited to Lobster parser logic and colocated tests, with no dependency, lockfile, CI, secret, permission, install, or supply-chain surface change.

Review details

Best possible solution:

Land one canonical Lobster parser fix with real installed-plugin or Gateway proof that ordinary default-field run/resume stays on the runner path while managed TaskFlow mode remains intact, then close the duplicate candidates and linked issue after merge.

Do we have a high-confidence way to reproduce the issue?

Yes, source-level. Current main and v2026.6.11 route default flow fields through the managed-parser decision path before ordinary runner fallback; I did not run a live Gateway reproduction in this read-only review.

Is this the best way to solve the issue?

Unclear as the final landing path. A plugin-local parser fix is the right owner boundary, but this branch is not clearly the best candidate while the sibling PR has stronger real-runtime proof.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 5a5b4c5b105f.

Label changes

Label justifications:

P2: The PR targets a normal-priority shipped Lobster plugin approval workflow bug with limited plugin-local blast radius.
rating: 🦪 silver shellfish: Overall readiness is 🦪 silver shellfish; proof is 🦪 silver shellfish and patch quality is 🐚 platinum hermit.
status: 📣 needs proof: The PR needs real behavior proof before ClawSweeper can clear the contributor ask. Needs real behavior proof before merge: The terminal proof exercises patched createLobsterTool with fake runner and TaskFlow dependencies; add redacted installed-plugin, Gateway, or real embedded-runtime output, then update the PR body to trigger a fresh ClawSweeper review or ask a maintainer to comment @clawsweeper re-review.

Evidence reviewed

PR surface:

Source +16, Tests +125. Total +141 across 2 files.

View PR surface stats

Area	Files	Added	Removed	Net
Source	1	22	6	+16
Tests	1	125	0	+125
Docs	0	0	0	0
Config	0	0	0	0
Generated	0	0	0	0
Other	0	0	0	0
Total	2	147	6	+141

What I checked:

Repository policy read: Root AGENTS.md and scoped extensions/AGENTS.md were read fully; plugin-boundary, compatibility, duplicate-candidate, and proof guidance affected this review. (AGENTS.md:1, 5a5b4c5b105f)
Current-main failure path: Current main parses flowStateJson and treats any flowExpectedRevision as a managed-mode signal before the ordinary runner fallback, matching the reported schema-default misrouting shape. (extensions/lobster/src/lobster-tool.ts:112, 5a5b4c5b105f)
Latest release still affected: The latest release tag v2026.6.11 contains the same flowStateJson and flowExpectedRevision managed-mode checks, so the linked bug is shipped behavior rather than only current-main churn. (extensions/lobster/src/lobster-tool.ts:111, e085fa1a3ffd)
Documented ordinary and managed contracts: The Lobster docs show ordinary run and resume payloads without flow fields, while managed TaskFlow mode is documented as the separate flowControllerId/flowGoal or flowId/flowExpectedRevision path. Public docs: docs/tools/lobster.md. (docs/tools/lobster.md:254, 5a5b4c5b105f)
PR parser behavior: At PR head, blank flow state is absent, empty-object flow state no longer selects run mode by itself, flowExpectedRevision: 0 no longer blocks ordinary calls, and explicit managed run still preserves stateJson: {}. (extensions/lobster/src/lobster-tool.ts:97, 6f2177f2ac8a)
PR regression coverage: The PR adds focused tests for ordinary run and resume with injected default flow fields, non-default revision without flowId, and explicit empty state in managed run mode. (extensions/lobster/src/lobster-tool.test.ts:110, 6f2177f2ac8a)

Likely related people:

mbelinky: Related merged PRs show this author added managed TaskFlow mode and the follow-up Lobster hardening that shaped the parser branch involved here. (role: introduced behavior and feature-history owner; confidence: high; commits: 30dc24fbd8be, b167df78aa17; files: extensions/lobster/src/lobster-tool.ts, extensions/lobster/src/lobster-tool.test.ts, extensions/lobster/src/lobster-taskflow.ts)
vincentkoc: Current parser blame points to e7365d41, recent history includes Lobster/check typing work, and the live PR is assigned to this account. (role: recent area contributor and assigned reviewer; confidence: high; commits: e7365d41d356, 87bcfe796fe4; files: extensions/lobster/src/lobster-tool.ts, extensions/lobster/src/lobster-tool.test.ts)
steipete: Recent history shows managed-flow formatting and tool-handling refactors near the same parser and TaskFlow result surface. (role: adjacent refactor contributor; confidence: medium; commits: 3700f3a22cd1, 95df6d933282; files: extensions/lobster/src/lobster-tool.ts, extensions/lobster/src/lobster-taskflow.ts)
kirkluokun: This contributor changed the approvalId/resume-token behavior in the same Lobster approval resume path. (role: adjacent resume contributor; confidence: medium; commits: 75fc8b704027; files: extensions/lobster/src/lobster-tool.ts, extensions/lobster/src/lobster-runner.ts)

What the crustacean ranks mean

🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

How this review workflow works

ClawSweeper keeps one durable marker-backed review comment per issue or PR.
Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
Maintainers can also comment @clawsweeper review to request a fresh review only.
Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

Review history (4 earlier review cycles)

reviewed 2026-07-08T05:04:27.124Z sha fcc7cc9 :: needs real behavior proof before merge. :: none
reviewed 2026-07-08T05:11:56.628Z sha fcc7cc9 :: needs real behavior proof before merge. :: none
reviewed 2026-07-08T05:26:57.718Z sha cd9cb4d :: needs real behavior proof before merge. :: none
reviewed 2026-07-08T05:34:40.270Z sha cd9cb4d :: needs real behavior proof before merge. :: none

LiLan0125 · 2026-07-08T06:03:40Z

@clawsweeper re-review

Updated the proof section with a real embedded runtime run: the patched Lobster tool loaded the installed @clawdbot/lobster package, ran a temporary .lobster approval workflow with flowStateJson: "{}" and flowExpectedRevision: 0, then resumed it with the returned real token and the same default flow fields.

Verified locally on PR head c12bf2f:

pnpm test extensions/lobster/src/lobster-tool.test.ts extensions/lobster/src/lobster-runner.test.ts: passed
node --import tsx --no-warnings embedded-runtime proof: runStatus needs_approval, resumeStatus ok

clawsweeper · 2026-07-08T06:03:43Z

🦞👀
ClawSweeper picked this up.

Command router queued. I will update this comment with the next step.

vincentkoc · 2026-07-08T06:03:47Z

Maintainer validation is complete on exact head c12bf2f48570da79968487e1dcc10fb4b9ac2f35.

The final parser distinguishes schema-injected defaults from intentional managed fields while preserving strict cross-mode validation. I added the missing negative coverage after fresh autoreview caught two cases where incompatible fields could otherwise be dropped before fallback.

Proof:

AWS Crabbox run_f48d59554de8 on lease cbx_c92971815ef7: exit 0; lease released.
52 focused Lobster tool/runner tests passed.
Real embedded run -> approval -> resume with injected {} and revision 0 passed.
Real managed run preserved explicit stateJson: {}.
Full remote pnpm check:changed passed.
Hosted CI run 28920766065 completed with 63 successful checks and no failures.
Fresh autoreview is clean.

Ready to squash merge.

vincentkoc · 2026-07-08T06:04:08Z

Merged via squash.

Prepared head SHA: c12bf2f48570da79968487e1dcc10fb4b9ac2f35
Landed commit: 8cb92636920b02306f9b50fa82fb4b1313a0237f

LiLan0125 · 2026-07-08T06:05:48Z

@clawsweeper hatch

…aw#102036) * fix(lobster): keep default flow fields on runner path * fix(lobster): reject incomplete managed resume revisions * fix(lobster): preserve explicit empty managed flow state * fix(lobster): preserve cross-mode validation --------- Co-authored-by: Vincent Koc <[email protected]>

fix(lobster): keep default flow fields on runner path

fcc7cc9

openclaw-barnacle Bot added extensions: lobster Extension: lobster size: S labels Jul 8, 2026

clawsweeper Bot mentioned this pull request Jul 8, 2026

fix(lobster): don't route schema-default flowExpectedRevision into TaskFlow (#102011) #102022

Closed

fix(lobster): reject incomplete managed resume revisions

cd9cb4d

clawsweeper Bot mentioned this pull request Jul 8, 2026

fix(lobster): treat schema-default flowExpectedRevision=0 as absent for ordinary run/resume #102023

Closed

1 task

vincentkoc self-assigned this Jul 8, 2026

LiLan0125 and others added 2 commits July 8, 2026 13:42

fix(lobster): preserve explicit empty managed flow state

6f2177f

fix(lobster): preserve cross-mode validation

c12bf2f

openclaw-barnacle Bot added size: M and removed size: S labels Jul 8, 2026

vincentkoc merged commit 8cb9263 into openclaw:main Jul 8, 2026
112 of 115 checks passed

vincentkoc mentioned this pull request Jul 8, 2026

fix(lobster): treat schema-default flowExpectedRevision=0 as absent for ordinary run/resume #102043

Closed

1 task

github-actions Bot mentioned this pull request Jul 8, 2026

📡 Upstream Digest — 2026-07-08 08:10 UTC curtismercier/openclaw-mods#1237

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

fix(lobster): keep ordinary run/resume on default flow fields#102036

fix(lobster): keep ordinary run/resume on default flow fields#102036
vincentkoc merged 4 commits into
openclaw:mainfrom
LiLan0125:fix/102011-lobster-flow-defaults

LiLan0125 commented Jul 8, 2026 •

edited by vincentkoc

Loading

Uh oh!

clawsweeper Bot commented Jul 8, 2026 •

edited

Loading

Uh oh!

LiLan0125 commented Jul 8, 2026

Uh oh!

clawsweeper Bot commented Jul 8, 2026

Uh oh!

vincentkoc commented Jul 8, 2026

Uh oh!

Uh oh!

vincentkoc commented Jul 8, 2026

Uh oh!

LiLan0125 commented Jul 8, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Uh oh!

Conversation

LiLan0125 commented Jul 8, 2026 • edited by vincentkoc Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What Problem This Solves

Why This Change Was Made

User Impact

Evidence

Uh oh!

clawsweeper Bot commented Jul 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

LiLan0125 commented Jul 8, 2026

Uh oh!

clawsweeper Bot commented Jul 8, 2026

Uh oh!

vincentkoc commented Jul 8, 2026

Uh oh!

Uh oh!

vincentkoc commented Jul 8, 2026

Uh oh!

LiLan0125 commented Jul 8, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

LiLan0125 commented Jul 8, 2026 •

edited by vincentkoc

Loading

clawsweeper Bot commented Jul 8, 2026 •

edited

Loading