Skip to content

Clarify planning boundaries and share review continuation for 6.3.1 - #141

Merged
mzored merged 3 commits into
mainfrom
fix/plan-review-continuity
Oct 7, 2026
Merged

mzored merged 3 commits into
mainfrom
fix/plan-review-continuity

Conversation

@mzored

@mzored mzored commented Oct 7, 2026

Copy link
Copy Markdown
Owner

Summary

Planning now distinguishes grants for a future executor from an explicit request to execute after preparation. It stops at the prepared plan by default, respects an explicit plan-then-execute request, and accounts for any implementation already created. Workload assessment happens while drafting tasks and remains subject to independent review.

Plan and longrun share one correction-review loop in verification: preserve candidate identity, findings, dispositions and reviewer continuity; independently review material corrections and affected consequences; reuse applicable evidence; and report unfinished review as a readiness limit. No universal deadline, role roster, new module, or public interface is introduced. This is the 6.3.1 PATCH candidate.

Verification

  • Five original arm-aware synthetic scenarios added; existing task-sizing probes retained. The new fixture is an overlay, so historical fixture identities remain unchanged.
  • Full local python scripts/check.py: 384 passed, 22 failures confined to stale instrument version fields. After correction, 62 affected corpus and package/document checks passed. Unchanged full-run evidence is reused under AGENTS.md; the final documentation-only check also passed.
  • Clean git diff --check.
  • Isolated independent read-only Codex review found the base-host oracle prerequisite and stale versions. Both findings were confirmed and fixed. The same reviewer approved the targeted correction review; no findings were rejected.
  • Strict Claude schema validation passed locally. The pinned Codex validator runs in CI. Packaging and installation mechanisms are unchanged; clean-install smoke is not required for this prose-only package change.
  • No paid behavioral comparison was authorized or run. Behavior, reliability, speed and savings remain UNVERIFIED.

Notes

The decision, evidence, prior-art comparison, and changelog record the limited source diagnosis. No private session text, identifiers, or project data are included. VERSION, both manifests, instrument versions, and the evidence page are aligned. The release workflow publishes the per-capability validation matrix.

@mzored
mzored merged commit d0f702b into main Oct 7, 2026
1 check passed
@mzored
mzored deleted the fix/plan-review-continuity branch October 7, 2026 02:48
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