Skip to content

Prevent negated maintainer readiness from approving strategic issues - #193

Merged
laclance merged 3 commits into
devfrom
fix/negated-maintainer-readiness
Oct 9, 2026
Merged

laclance merged 3 commits into
devfrom
fix/negated-maintainer-readiness

Conversation

@laclance

@laclance laclance commented Oct 9, 2026

Copy link
Copy Markdown
Owner

Finding

The shared readiness detector considered the phrase ready for implementation an approval even when embedded in Not ready for implementation. Because positive readiness is evaluated before maintainer hold logic, an explicit refusal could be misread as permission, allowing a strategic proposal or upstream-tool issue into the queue.

Fix

  • Detect explicit local negations of readiness/implementation-welcome phrases, and never treat them as a positive readiness signal.
  • Classify these negated statements as an implementation hold, while preserving the existing latest-trusted-maintainer-comment-wins rule.
  • Reuse the shared decision in maintainer-authored issue bodies and both recently added strategic-readiness filters; no new API requests or scanner flow changes.
  • Add unit tests for negated variants, genuine approvals, comment ordering/authority, and end-to-end strategic rejection. Document the rule.

Scope

Narrow readiness correctness fix. No other audit findings addressed, no private-instance deployment, no changes to paid, discovery, scoring, state persistence, or delivery.

Validation

Requires green CI formatting/lint, strict types, coverage and Python compatibility. Remains a draft PR until explicitly authorized for merge.

@laclance
laclance marked this pull request as ready for review October 9, 2026 11:16
@laclance
laclance merged commit eaf0175 into dev Oct 9, 2026
4 checks passed
laclance added a commit that referenced this pull request Oct 9, 2026
Release promotion of five reviewed fixes: durable recovery output formatting; upstream Go-tool attribution; architecture proposal approval; negated maintainer readiness; and proxy-header feature versus existing-regression classification.

Includes #190, #191, #192, #193 and #194. Preserve dev squash-only history.
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