Repository navigation
feat(session-flow): one stage order with Interview first, decision levels and the handoff threshold #12663
Description
Parent
Source: PLAN Phase 7, topic planning-chain-consolidation. Epic: #12646.
Depends on
Outcome
The workflow lists one stage order, interview first, with the stage named Interview, and its list and prose agree; the interview, design and plan skills each state the level they decide; between human stages the workflow suggests a handoff only at context-guard zone dumb.
Demo: Read the workflow skill's stage list and prose: both start at Interview and agree.
The Brief (Problem, Goal, Constraints) is in the epic body.
Decisions
- Q4: Behaviour is decided in the Brief, interfaces and contracts in /planning:design, files and symbols in the plan.
- Q23: Human stages run in one session; a handoff is suggested only when context-guard warns.
- Q42: A handoff is suggested only at context-guard zone dumb.
Key interfaces
Done when
- AC1: the workflow's stage list and prose agree, interview first; no stage named Contract.
- AC2: the interview, design and plan skills each carry one sentence naming the level they decide.
- AC12: the workflow suggests a handoff between human stages only at zone dumb (grep shows the rule line).
- Before editing, check ListAgents and open PRs on the workflow skill (an open PR edits it) and coordinate.
Out of scope
- Vocabulary fixes beyond the stage name (vocabulary slice)
Metadata
Metadata
Assignees
Labels
No labels
Type
Fields
No fields configured for Feature.
Activity
github-actions commentedon Nov 3, 2026
melodic-automation-lanes commentedon Nov 3, 2026
This records a re-fire of an earlier refusal, not a new refusal. The lanes kill switch stopped run 41576123623, the work-resolve-issue run that
agent-readystarted, before it did any work; that run ended at 2026-11-03T04:45:53Z. work-unblock has now removed and addedagent-readyback to start a fresh run. A kill-switch refusal is no fault in this issue or the lanes, so it never counts against later re-fires; after limits.limit-stops-per-issue of them since a person last addedagent-readyor a run last reported, work-unblock waits longer before each one, up to 24 hours.github-actions commentedon Nov 3, 2026
melodic-automation-lanes commentedon Nov 3, 2026
This records a re-fire of an earlier refusal, not a new refusal. The lanes kill switch stopped run 41582468768, the work-resolve-issue run that
agent-readystarted, before it did any work; that run ended at 2026-11-03T05:45:16Z. work-unblock has now removed and addedagent-readyback to start a fresh run. A kill-switch refusal is no fault in this issue or the lanes, so it never counts against later re-fires; after limits.limit-stops-per-issue of them since a person last addedagent-readyor a run last reported, work-unblock waits longer before each one, up to 24 hours.github-actions commentedon Nov 3, 2026
github-actions commentedon Nov 3, 2026
kyle-sexton commentedon Nov 3, 2026
Claimed by the planning-chain consolidation session (HTML Plan): finishing PR #12787 by merging main (S6 #12785 landed on the same interview skill) and adding tripwire tests for AC2 and AC12. Its review lanes failed only during the review-token outage and will re-run on the new head.