security: remove operational runbooks from PUBLIC molecule-core #3183
Reference in New Issue
Block a user
Delete Branch "security/remove-public-runbooks-20260623"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
6 operational runbooks were unauthenticated-readable in this PUBLIC repo (production-auto-deploy, sop-production-cicd, gitea-merge-queue, gitea-operational-quirks, gitea-actions-migration-checklist, dev-sop). Canonical copies live in the private internal repo; internal PR #937 moves the 3 that were not already there. This removes runbooks/ from HEAD. NOTE: git HISTORY still contains them (public) — separate CTO decision on history-scrub vs accept-and-rotate. Root cause: they entered via PR #881 (396 files mislabeled '36 test cases') + #2159; review-discipline SOP added in internal PR #937. Expedite merge — this reduces live exposure.
History decision (CTO, 2026-06-23): ACCEPT-AND-NOTE — no scrub. Secret-value scan of all 6 runbooks returned 0 live secrets (no AWS keys, GH/JWT tokens, embedded creds, or high-entropy values). Exposed content is infra identifiers + procedures only; treated as known-to-adversaries and defended by IP-allowlists/IAM, not by secrecy. Public-history rewrite declined: high disruption (force-push breaks clones/forks/open PRs) and ~6 weeks public = likely already harvested, so low real benefit. This PR removes them from HEAD; git history retains them by design decision.
APPROVED on
05f70b48.5-axis review: this removes the public runbooks directory from HEAD and narrows live exposure without pretending git history is scrubbed; the PR body correctly leaves history-scrub/rotation as a separate CTO decision. Correctness/security: the six operational runbooks are deleted, and a PR-head tree search found no stale references to those deleted public runbook paths; remaining hits are script/workflow names or references to private/internal runbooks. Robustness/performance: deletion-only, no runtime impact. Readability: scope is clear and tightly limited.
5-axis review: Correctness: the PR deletes the public operational runbooks and I found no remaining references to the deleted runbook paths on this head. Security: this reduces public exposure; the residual git-history exposure is explicitly acknowledged as a separate CTO decision. Robustness/readability: clean deletion-only change. Performance: no impact. No blocking findings; pending E2E/governance contexts are separate from the review verdict.