Skip to content

fix(code-review): revert #178 parallel GitHub fleet → restore synchronous headless dispatch (FEA-3085) - #179

Merged
peterulsteen merged 2 commits into
mainfrom
fix/code-review-github-sync-fleet
Jul 14, 2026
Merged

fix(code-review): revert #178 parallel GitHub fleet → restore synchronous headless dispatch (FEA-3085)#179
peterulsteen merged 2 commits into
mainfrom
fix/code-review-github-sync-fleet

Conversation

@peterulsteen

Copy link
Copy Markdown
Contributor

Summary

Reverts #178 (code-review v3.5.0) and bumps 3.4.1 → 3.6.0, restoring synchronous reviewer-fleet dispatch for GitHub/headless mode at stage_20_spawn_reviewers. Local mode keeps parallel background dispatch + blocking TaskOutput.

#178 unified both modes on run_in_background: true + blocking TaskOutput, betting that headless claude -p (Claude Code ≥ v2.1.182) reliably awaits background subagents as long as the orchestrator issues the blocking collection. Empirically it does not: in ~18% of symphony-alpha claude-code-review.yml pull_request runs on 2026-07-13 (9/50, up from 1/10 the day #178 shipped) the orchestrator launched the reviewer fleet (Bug Hunter A/B, Unified Auditor) as background agents and then ended its turn ("I'll wait for the harness to notify me") with no pending synchronous tool call. claude -p then exited (terminal_reason: "completed") before stage_21_collect_findings … stage_30_footer ran — no findings collected, no .closedloop-ai/code-review-* artifacts written, and the workflow's "Validate migrated review artifacts" guard failed the job. Runs that took the synchronous fast-path were unaffected, hence the intermittency.

Synchronous one-at-a-time dispatch (run_in_background: false, wait for each DONE findings=N file=... before the next) keeps the headless turn alive by construction, rather than depending on the model choosing to issue a blocking TaskOutput. This restores the exact pre-#178 state that ran reliably (the verifier fleet at stage_23 was already background+collect in both modes before #178 and is unchanged).

Trade-off: GitHub review wall-clock returns to sum-of-reviewers instead of slowest-reviewer. Correctness over speed for a merge-adjacent gate.

Root-cause detail and the 9 failing run IDs are tracked in FEA-3085.

Feature Flags

N/A — plugin behavior change, no runtime flag.

Breaking changes

None. Restores prior GitHub-mode dispatch semantics; output contract and artifacts unchanged.

Test plan

  • uv run --frozen --group dev pytest2079 passed, 3 skipped
  • ruff check . → clean
  • pyright → 0 errors, 0 warnings
  • Restored FEA-2162 synchronous-GitHub reviewer test class passes.
  • Post-merge: watch symphony-alpha claude-code-review.yml pull_request runs for the artifact-validation failure signature (expect it to clear once CI pulls v3.6.0).

🤖 Generated with Claude Code

…ous headless dispatch (FEA-3085)

Reverts d7362f0 (v3.5.0, #178) and bumps 3.4.1 -> 3.6.0.

#178 unified GitHub and local reviewer-fleet dispatch (stage_20) on
run_in_background: true + blocking TaskOutput, on the premise that
headless `claude -p` (Claude Code >= v2.1.182) reliably awaits background
subagents as long as the orchestrator issues the blocking collection.

Empirically the orchestrator does NOT reliably issue it. In ~18% of
symphony-alpha claude-code-review.yml pull_request runs on 2026-07-13
(9/50, up from 1/10 the day #178 went live) it launched the fleet
(Bug Hunter A/B, Unified Auditor) as background agents and then ended its
turn ('I'll wait for the harness to notify me') with no pending
synchronous tool call, so claude -p exited (terminal_reason: completed)
before stage_21_collect_findings..stage_30_footer ran. No findings were
collected, no .closedloop-ai/code-review-* artifacts were written, and the
workflow's 'Validate migrated review artifacts' guard failed the job.
Runs that took the synchronous fast-path branch were unaffected, which is
why the failure was intermittent.

Restore the known-good split: GitHub/headless standard flow dispatches
reviewers synchronously one-at-a-time (run_in_background: false, wait for
each 'DONE findings=N file=...' before the next), which keeps the turn
alive by construction rather than depending on the model choosing to
issue a blocking TaskOutput; local mode keeps parallel background dispatch
+ blocking TaskOutput. Restores the FEA-2162 synchronous-GitHub reviewer
contract and its test class. Trade-off: GitHub review wall-clock returns
to sum-of-reviewers -- correctness over speed for a merge-adjacent gate.

Full code-review suite green (2079 passed, 3 skipped), ruff + pyright clean.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

@mikeangstadt mikeangstadt left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sound revert. Restores the sync GitHub dispatch that we know keeps the headless turn alive, and the version bump + tests line up. One real question about whether the verifier fleet has the same headless death you're fixing here, plus a changelog nit. Not blocking either.

Comment thread plugins/code-review/skills/verify-findings/SKILL.md
Comment thread CHANGELOG.md
…t (not replace)

Address review: v3.5.0 shipped to main, so its changelog block must stay
(entries are newest-first + treated as released on merge). Put v3.6.0 on
top of the retained v3.5.0 Changed block instead of deleting it. Also note
in v3.6.0 that stage_23 verifiers are unchanged and their headless
exposure is tracked in FEA-3154.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@peterulsteen
peterulsteen merged commit 1f07321 into main Jul 14, 2026
5 checks passed
@peterulsteen
peterulsteen deleted the fix/code-review-github-sync-fleet branch July 14, 2026 19:28
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants