docs: correct stale comments left over from prior refactors

[sumitake]

Summary

Reapplies 3 of 11 uncommitted local edits that were stashed before pulling 123 commits from origin/main. The other 8 were checked against the new upstream content and dropped: 3 were superseded by upstream changes (schema version bump, a deleted/consolidated skill file, a rewritten migration doc), 4 were based on a misunderstanding ({{ mcp_tool_ask }} is a live build-time template variable resolved by scripts/skill-build-config.json, not stale text — hardcoding it would have broken per-package templating), and 1 (check-public-export-safety.py's AST-based literal-masking behavior) needs deeper review than a comment-fix PR should carry, so it's being left out rather than guessed at.

• scripts/build-changelog.py: the module-level DESIGN comment still described the old positional-insertion strategy (insert BETWEEN [Unreleased] and the first ### heading). The compiler has used boundary-marker replacement (FRAGMENT_BLOCK_START/END) for a while now, and compile_changelog()'s own docstring already documents that correctly — only the top-of-file comment was stale.
• scripts/build_skills.py: PACKAGES_FIELD_RE only matches the inline bracket form (packages: [a, b]); the comment claiming YAML block-list support was never accurate for this regex. Also drops --package's help text reference to an _active filter that resolve_packages() no longer has (there's now exactly one supported package, enforced by SystemExit).
• docs/superpowers/plans/2026-07-12-agent-collab-v3.1.0-license.md: fixes a script-name typo (build_changelog.py → build-changelog.py) in a verification command block.

No behavior change — see Verification.

Boundary declaration

• No provider executor source, raw provider command, credential, private absolute path, retired package tree, downloader, or post-install hook is included.
• Native-runtime changes, if any, contain only a final signed artifact and reviewed public verification metadata; implementation and credentials remain private. (N/A — no runtime changes.)
• The change does not create a host-specific preset or provider-specific plugin.

Generated and release surfaces

• Skill specs and generated SKILL.md files are in parity. (build_skills.py --check passes identically before/after.)
• Claude and Codex marketplaces/manifests are in parity. (Unaffected — no manifest/version changes.)
• A unique changelog.d/ fragment — not added; this is a comment/typo-only fix with no user-visible behavior change.
• Version metadata bump — not applicable; no distributed content or behavior changed.

Verification

• python3 scripts/build_skills.py --check → OK: all generated SKILL.md files match source specs.
• python3 -m unittest scripts.test_build_changelog tests.test_build_skills_description_limit -v → 24 tests, all pass.
• git diff --check → clean.
• python3 scripts/check_release_consistency.py (ran via pre-commit hook) → RESULT: OK.
• python3 scripts/build-changelog.py --check — pre-existing, unrelated failure (uncompiled fragment drift on main, not caused by this change).
• Full python3 -m unittest discover / secret_scan.py / check-public-export-safety.py --active-tree --history — blocked in this local checkout by pre-existing untracked debris unrelated to this diff: stray retired package directories under plugins/ (antigravity-collab, claude-collab, codex-collab — confirmed untracked via git ls-files, not nested repos, left over from before the provider-package retirement) and an unreadable .claude-worktrees/ directory. Scoped the two dedicated unit-test modules above instead, which cover the exact functions touched.

Review and post-condition

Tier: 1 (pure comment/typo correction; zero logic change — confirmed by build_skills.py --check passing identically and by reading compile_changelog()'s own docstring, which already matches current behavior). No cross-check round required per Tier 1 process.

Post-condition: none — no runtime, schema, or generated-artifact change to verify post-merge.

Compliance trace

author: claude
standing_directives: agent-collab-workspace directive #6 (Tier 1 fast-path); Session calibration (verify by positive observation)
tier: 1
cross_check: N/A — Tier 1 pure comment/typo fix, no cross-check round required
post_condition: none — no runtime/schema/generated-artifact change
mcp_coverage_gap: NONE
contributor_rights: OPERATOR-CONFIRMED
operator_reserved: no

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Advanced

Run ID: ea985f30-3d97-4c52-96f5-c1a38cd59ac4

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-25T03:49:59.069197Z	431e27d	PR opened
🔒 Security Review	✅ Completed 2026-09-25T03:52:08.262377Z	431e27d	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.