docs(skills): Contents blocks on long reference spokes (check 26) (#4071) - #4781
Conversation
|
PR body contract — issue linkage This PR body conforms to the issue-linkage contract. Nothing to do. |
ae98837 to
831936b
Compare
…lags (#4071) Co-authored-by: Kyle Sexton <kyle-sexton@users.noreply.github.com>
831936b to
b2cbf04
Compare
The contents-block notes move to claude-config 0.51.23, claude-ops 0.63.23, and source-control 0.62.10. Main's 0.51.19, 0.51.18, 0.51.17, 0.63.17, 0.63.16, and 0.62.6 sections stay. Co-authored-by: Kyle Sexton <kyle-sexton@users.noreply.github.com>
markdownlint MD012 flags a trailing blank line as two consecutive blanks. Co-authored-by: Kyle Sexton <kyle-sexton@users.noreply.github.com>
# Conflicts: # plugins/claude-config/.claude-plugin/plugin.json # plugins/claude-config/CHANGELOG.md # plugins/claude-ops/.claude-plugin/plugin.json # plugins/claude-ops/CHANGELOG.md # plugins/source-control/.claude-plugin/plugin.json # plugins/source-control/CHANGELOG.md Co-authored-by: Kyle Sexton <kyle-sexton@users.noreply.github.com>
# Conflicts: # plugins/claude-config/.claude-plugin/plugin.json # plugins/claude-config/CHANGELOG.md # plugins/claude-ops/.claude-plugin/plugin.json # plugins/claude-ops/CHANGELOG.md # plugins/claude-ops/skills/audit-install-state/SKILL.md # plugins/claude-ops/skills/audit-skill-visibility/SKILL.md Co-authored-by: Kyle Sexton <kyle-sexton@users.noreply.github.com>
…criptions Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PT4esxbdC7eQiwQxy35Mie
|
Claude finished @kyle-sexton's task in 1m 19s —— View job Security review complete — no issues found
I read the full diff (
None of this touches code paths that process untrusted input, authentication/authorization, secrets/credentials, injection surfaces (command/SQL/path/template), or GitHub Actions workflow configuration. The Skip gate applied: no security-relevant surface in this diff. No findings to report. |
|
Claude finished @kyle-sexton's task in 3m 49s —— View job Code review complete
Ran Verification method: for each of the ten Contents blocks and the two new headings in Finding (posted inline on Not flagged, but worth a human glance: One informational note: the PR body's listed patch-bump versions (e.g. "claude-config 0.50.3") no longer match the diff (claude-config is actually 0.51.21 → 0.51.23) — expected drift from stacking on #4112/main, not a code issue. |
Resolve CHANGELOG.md/plugin.json conflicts in claude-ops, source-control (PR version already above main's) and code-tidying (identical-bump collision at 0.23.11, renumbered to 0.23.12). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PT4esxbdC7eQiwQxy35Mie
|
Claude finished @kyle-sexton's task in 45s —— View job Security review complete — no issues found
I read the full diff (
Skip gate applied: no security-relevant surface in this diff — no untrusted input handling, auth/authorization, secrets/credentials, injection surfaces (command/SQL/path/template), or GitHub Actions workflow changes. The |
…4112) (#4767) <!-- CURSOR_AGENT_PR_BODY_BEGIN --> Closes #4112 ## Summary The native-surface presence gate was `resolves in your session`, which addresses the reader. #4108 removed second person from every other description clause but had to keep this one: `audit-native-overlap`'s registry self-check matches the literal token. This PR changes the token everywhere in one coordinated change to `resolves in this session`. The suggest token `is available in your session (` from native-references 2.0.0 is unchanged. ## Fix - `overlap.py` `GATE_TOKEN` is `resolves in this session`. Forward/reverse parity and ungated-presence advisory follow the constant. - Bodies, references, evals, docs: every occurrence of the route token outside historical CHANGELOGs. The native-references README names `resolves in your session` as the retired spelling that is no longer accepted. - `docs/conventions/native-references` major 3.0.0 for the canonical-token change. - Patch bumps: claude-config 0.51.22, claude-ops 0.63.25, code-tidying 0.23.11, evals 0.3.2, playbooks 0.13.23, prototype 0.13.1, review 0.33.1, source-control 0.62.9, testing 0.9.5, visualization 0.8.1. ## Verification - `test_overlap.py`: 117 passed; `overlap.test.sh` exit 0; registry self-check degraded (stale-but-honest advisories only), never broken. - `validate-plugins.sh`: all manifests and the catalog validated. `check-changed-skills.sh origin/main`: 15 skills, 0 failed. - `check-changelog-parity.sh --check` and `--check-order`, `check-purged-em-dashes.sh`, `check-docs-naming.sh --check`, markdownlint on the native-references README: ok. - `resolves in your session` appears outside CHANGELOGs only in the README line that names it as retired. ## Related - #4108, #4069 / #4072. Stacked on by #4071 (#4781). ### Decision record - **Claim:** A skill description must be written in the third person because it is injected into the system prompt, so the gate token cannot say "your". - **Basis:** [Skill authoring best practices, "Writing effective descriptions"](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/best-practices#writing-effective-descriptions): always write in third person. Fetched 2026-09-28. - **As of:** 2026-09-28. - **Recheck:** when that page's description guidance changes, or when Claude Code starts reading descriptions somewhere other than the system prompt. <!-- CURSOR_AGENT_PR_BODY_END --> 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01Hx63mvV8t39y644DQ3uLnn --------- Co-authored-by: Cursor Agent <cursoragent@cursor.com> Co-authored-by: Kyle Sexton <kyle-sexton@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Resolve the merge by taking main for every file the PR touched, then re-apply only the Contents blocks on the ten long reference spokes and bump the seven owning plugins one patch above main. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WejvGhkHUSWPRim2caW2Wh
…tents-37e9 # Conflicts: # plugins/claude-ops/CHANGELOG.md # plugins/source-control/CHANGELOG.md
|
Claude finished @kyle-sexton's task in 38s —— View job Security review complete — no issues found
Reviewed the current PR head ( This PR's actual content:
None of this touches untrusted-input handling, authentication/authorization, secrets/credentials, injection surfaces (command/SQL/path/template), or GitHub Actions workflow files. No Skip gate applied: no security-relevant surface in this diff. No findings to report. |
|
Claude finished @kyle-sexton's task in 1m 59s —— View job Code review complete
Scope: ten Verification method: for each of the ten Contents blocks (and the two new H2s in For Version bumps and CHANGELOG entries are consistent: each of the 7 touched plugins bumps exactly one patch version above its current No findings. This is a mechanical, low-risk change (docs navigation only, no content moved or altered) and it's internally consistent. I could not execute |
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WejvGhkHUSWPRim2caW2Wh
|
Claude finished @kyle-sexton's task in 1m 41s —— View job Security review complete — no issues found
Reviewed the current PR head (
None of this touches untrusted-input handling, authentication/authorization, secrets/credentials, injection surfaces (command/SQL/path/template), or GitHub Actions workflow files — no Skip gate applied: no security-relevant surface in this diff. No findings to report. |
|
Claude finished @kyle-sexton's task in 2m 10s —— View job Code review complete
Scope: Verification method: I read each of the ten touched files in full at the current head (not just the diff hunks) and, for every entry in its new Version bumps and CHANGELOG entries are internally consistent: each of the 7 touched plugins bumps exactly one patch version above its pre-PR value, with a changelog entry naming the specific spoke(s) touched. On the earlier review threads on this branch: the two prior code-review comments here were against an older head ( No findings. This is a mechanical, low-risk, docs-only navigation change with no content moved or altered, and it holds up under a full anchor-by-anchor check rather than a sampled one. Not independently verified: |
Closes #4071
Summary
Ten long reference spokes (over 300 lines) open with a
## Contentsblock of section anchors, soskill-quality check 26 (long spoke files carry a table of contents) passes on them.
Fix
audit-pass/reference/determinism-tiers.md,audit-pass/reference/run-state-and-resumability.md(adds the lease sub-anchor to its existing block),audit/context/validation-categories.mdaudit/reference/official-guidance.mdaudit-performance/reference/known-performance-issues.md,observability/context/data-sources.mdresearch/context/dispatch.mdaudit/context/surface-walk.mdsetup/reference/apply-convention.mdgains two headings (target layer and non-interactive writes; the interactive interview) so its Contents block has anchors to link; no step text changedbe-concise/reference/sources.mdEach of the seven plugins is bumped one patch above main with a dated CHANGELOG entry. No other file changes.
Verification
scripts/check-changed-skills.sh origin/main: every touched skill PASS, no check 26 warning on any of the ten spokesscripts/validate-plugins.sh: all manifests and the catalog validatescripts/check-changelog-parity.sh --check --check-order: passRelated
needs-humanwith an open question about C3 ratification; this PR adds navigation only and records no decision on that question.🤖 Generated with Claude Code
https://claude.ai/code/session_01WejvGhkHUSWPRim2caW2Wh