docs(docs-hygiene): clarify audit-derivability Hard Rules line 111 fork-mechanism reference

[kyle-sexton]

Summary

The audit-derivability Hard Rules bullet read "…never confirmed from this (contaminated) context, never by a fork." That bare "a fork" was technically correct — it means the Agent tool's fork subagent type — but a reader landing in the Hard Rules section first (via search or a partial read) had no local disambiguation and could wonder whether it also bans a skill's own context: fork frontmatter. This makes the bullet self-contained.

Fix

Name the mechanism inline and distinguish it from skill-level context: fork, matching the phrasing already established in the spot-test bullet (added in #1058), so the two clarifications read as one consistent voice:

…never by the Agent tool's fork subagent type (a skill's own context: fork frontmatter is unrelated — it starts blank).

Verification

Before:

• Load-bearing or contested deletions are spot-tested by a fresh, non-fork subagent — never confirmed from this (contaminated) context, never by a fork.

After:

• Load-bearing or contested deletions are spot-tested by a fresh, non-fork subagent — never confirmed from this (contaminated) context, never by the Agent tool's fork subagent type (a skill's own context: fork frontmatter is unrelated — it starts blank).

The Hard Rules bullet now carries the same Agent-tool-fork-vs-context: fork disambiguation as the spot-test bullet, so it no longer depends on the reader having reached that earlier line. Patch version bump (0.8.5 → 0.8.6) with a matching CHANGELOG entry.

Closes #1062

Related

• docs(docs-hygiene): audit-derivability SKILL.md Hard Rules line 111 'never by a fork' reads ambiguous without line 56's context #1062 — the deferred follow-up this addresses.
• docs: audit-derivability rubric/evals invert the context-fork inheritance claim #1053 — the original inverted-fork-mechanism issue.
• docs(docs-hygiene): fix inverted fork-mechanism attribution in audit-derivability rubric+evals #1058 — the fix that added the spot-test bullet's disambiguation this now mirrors in the Hard Rules section.
• docs(docs-hygiene): audit-derivability fork-mechanism wording drift (line 56 vs 111) + Gotchas line ~119 still undisambiguated #1082 — deferred follow-up (SUGGESTION, non-blocking): "opposite" vs "unrelated" wording drift between line 56 and line 111, plus the Gotchas bullet (~line 119) still lacking disambiguation.

---

Authored by Claude Code (Claude Opus 4.8). Session: https://claude.ai/code/session_01Mu8bLN896ia5AeWmGPqAUv

[claude]

Claude finished @kyle-sexton's task in 1m 3s —— View job

---

PR Review — docs(docs-hygiene): clarify audit-derivability Hard Rules line 111 fork-mechanism reference

• Gather context and read changed files
• Review SKILL.md clarification for correctness and consistency
• Review CHANGELOG.md entry
• Review plugin.json version bump

---

Summary

Tight, well-scoped doc clarification. All three files are consistent with each other and with the rest of SKILL.md. No functional changes — purely a documentation precision fix. Approved to merge.

---

SKILL.md — line 111

Before:

…never confirmed from this (contaminated) context, never by a fork.

After:

…never confirmed from this (contaminated) context, never by the Agent tool's fork subagent type (a skill's own context: fork frontmatter is unrelated — it starts blank).

The disambiguation is technically correct. One minor phrasing observation: the existing spot-test bullet at line 56 uses "the opposite" (a skill's own \context: fork` frontmatter is the opposite: it starts blank`), while the new Hard Rules bullet uses "unrelated". Both are accurate, but they aren't literally the same voice the PR description claims to mirror. This is a low-severity observation — "unrelated" is defensible and arguably more precise in the terse Hard Rules register.

One follow-up to consider (outside this PR's scope): the Gotchas bullet at line 119 still uses bare "non-fork subagent" without disambiguation. It's a prose summary rather than the definitive statement, so it's lower stakes, but it's the third site that could complete the trilogy.

CHANGELOG.md

Entry at [0.8.6] follows the established format (heading → ### Fixed → bullet), is clearly scoped, and accurately describes what changed and why. No issues.

plugin.json

Patch bump 0.8.5 → 0.8.6 is the correct semver classification for a documentation clarification with no behavioral change.

---

Overall: No blockers. The Hard Rules section is now self-contained, which is the stated goal. The minor "opposite" vs "unrelated" wording gap is worth tracking but not a reason to hold the PR.

[claude]

Claude finished @kyle-sexton's task in 34s —— View job

---

Security Review

• Gather context and read changed files
• Analyze changed files for security issues
• Post findings

---

No security issues found.

All three changed files are purely documentation and metadata:

• plugin.json — patch version bump (0.8.5 → 0.8.6), no executable content, no dependency changes, no permission changes.
• CHANGELOG.md — prose changelog entry, no executable content.
• SKILL.md — a single-line clarification to a Hard Rules bullet. The change adds explanatory prose; no code paths, no shell expansion surfaces, no secrets, no injection surfaces introduced.

There are no GitHub Actions workflow changes, no dependency pin changes, no permission widening, and no new remote resource references in this diff. Nothing here presents a security risk.

[kyle-sexton]

Post-green review pass (work-2 lane). CI fully green. No review threads to resolve (PR-level review/security-review reports only, both approve/no-findings).

Finding	Classification	Action
Overall fork-mechanism clarification at line 111	VALID, addressed	Approved by review bot
Security review	No findings	N/A
"opposite" (line 56) vs "unrelated" (line 111) wording drift	VALID, non-blocking (SUGGESTION-tier)	Deferred — filed #1082
Gotchas bullet (~line 119) still bare "non-fork subagent"	VALID, non-blocking (SUGGESTION-tier)	Deferred — filed #1082

Flipping this PR to ready-for-review and handing off to /source-control:babysit-prs for ongoing watch.

[claude]

Claude encountered an error after 0s —— View job

---

I'll analyze this and get back to you.

[claude]

Claude encountered an error after 0s —— View job

---

I'll analyze this and get back to you.

[github-actions]

Warning

Automated security review did not complete — this is an infrastructure failure, not a review verdict.

Treat any Claude comment on this PR (including a placeholder like "I'll analyze this and get back to you") as incomplete, not "no findings."

• Run: https://github.com/melodic-software/claude-code-plugins/actions/runs/29959600553
• Last SDK result: {"subtype":"success","is_error":true,"num_turns":1,"duration_ms":398,"total_cost_usd":0}

Re-running the job, or pushing a new commit, will retry the review.

[github-actions]

Warning

Automated review did not complete — this is an infrastructure failure, not a review verdict.

Treat any Claude comment on this PR (including a placeholder like "I'll analyze this and get back to you") as incomplete, not "no findings."

• Run: https://github.com/melodic-software/claude-code-plugins/actions/runs/29959600461
• Last SDK result: {"subtype":"success","is_error":true,"num_turns":1,"duration_ms":367,"total_cost_usd":0}

Re-running the job, or pushing a new commit, will retry the review.

[kyle-sexton]

🔒 babysit-prs lane claiming this PR for on-branch fix work this cycle (amendment-round: 16, safe tier). Will fix clear branch-owned findings and push; will not resolve threads or merge (safe tier).