Skip to content

docs: update consensus protocol for restart behavior [doc-updater] - #1102

Merged
jwbron merged 1 commit into
mainfrom
egg/doc-update-consensus-restart-docs
Mar 14, 2026
Merged

docs: update consensus protocol for restart behavior [doc-updater]#1102
jwbron merged 1 commit into
mainfrom
egg/doc-update-consensus-restart-docs

Conversation

@james-in-a-box

Copy link
Copy Markdown
Contributor

Update docs/guides/sdlc-pipeline.md to reflect the consensus wrapper behavior change from PR #1101 (Replace implicit consensus READY with agent restart).

Previously, item 6 in the Concurrent Execution Consensus Protocol section stated:

"If all containers exit before consensus, the phase completes based on container exit codes (same as non-concurrent mode)"

This described the old implicit-READY behavior where the orchestrator auto-registered READY for cleanly exited containers. That code has been removed. The new behavior is:

  • When a container exits cleanly (code 0) without signaling READY, the consensus wrapper inside the container restarts Claude with a recovery prompt
  • Restarts are capped at MAX_CONSENSUS_RESTARTS (default 2)
  • After exhausting restarts, the wrapper exits with code 1, triggering the HITL single-agent failure path

The docs/guides/concurrent-execution.md guide already has an accurate Consensus Wrapper section covering this. This PR updates sdlc-pipeline.md to match and links to the detailed guide.

Triggered by: #1101

Authored-by: egg

Update sdlc-pipeline.md to describe the new consensus wrapper behavior
introduced in PR #1101: instead of completing based on exit codes when
containers exit without signaling READY, the wrapper now restarts the
agent with a recovery prompt (up to MAX_CONSENSUS_RESTARTS, default 2).
After exhausting restarts, exit code 1 triggers the HITL failure path.

Authored-by: egg

@egg-reviewer egg-reviewer Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

No agent-mode design concerns. This is a documentation-only change updating the consensus protocol description to match the restart behavior implemented in #1101.

— Authored by egg

@egg-reviewer egg-reviewer Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Verified:

  • PR #1101 (the triggering code change) is merged
  • Cross-reference link concurrent-execution.md#consensus-wrapper resolves correctly to the ## Consensus Wrapper heading at line 48
  • MAX_CONSENSUS_RESTARTS default of 2 matches code in orchestrator/consensus_wrapper.py:19
  • New text accurately describes the per-container restart behavior: clean exit without READY → recovery prompt restart (capped) → exit code 1 → HITL single-agent failure path
  • Consistent with the detailed description in concurrent-execution.md lines 50-60

Clean documentation update. No issues.

— Authored by egg

@james-in-a-box

This comment has been minimized.

@james-in-a-box

Copy link
Copy Markdown
Contributor Author

egg review completed. View run logs

1 previous review(s) hidden.

@jwbron
jwbron merged commit 80e5bb1 into main Mar 14, 2026
33 checks passed
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.

1 participant