AI Development Methodology

E10 — External baseline audit remediation — Active Backlog

Items currently in scope for this epic. See charter for exit criteria.

Epic status (as of 2026-08-20): active, no open items. All fourteen chartered items are closed — see ARCHIVE.md. Every finding from the external audit is addressed: the Critical, all five High, and the four Medium. Staged for a single v1.32.0.

Two decisions shaped the shape of it. A committed integrity checker and any new CI were declined — this repo ships no runnable elements — so F-08 closes as a reproducible written procedure rather than a control, and the cost is stated in the charter instead of hidden. The cross-AI gate was removed from this epic’s exit criteria.

Summary

ID	Title	Priority	Effort	Status
—	(no open items — all fourteen archived)	—	—	—

What stands between this and closing

The epic stays active because two exit criteria are not things a working session can tick:

1. The release has not happened. Nothing is committed. The seven version-pin sites still read v1.31.0 and move in the release commit, not before — per the ordering rule this epic itself added.
2. The closing gate is a fresh cold re-audit against the final tree, returning no Critical and no High. It must be run by a session that did not author these fixes; this session cannot satisfy it, and claiming otherwise would be the exact defect the epic repaired.

Two maintainer decisions carried out of the sweep

Recorded in the convention sweep, not duplicated here. Neither blocks the release:

• Retire the house-verbosity setting? No mechanism, no user, no evidence — but retiring it is a removed section and therefore a MAJOR bump under this repo’s own SemVer table.
• Should the root AGENTS.md carry the context-integrity canary? The sweep recommends yes: the alternative is a repository publishing a safety convention it declined to use, which is the shape of the finding it just fixed.

This site is open source. Improve this page.