[kyle-sexton]

No related issue: the eight fix-route rows of the D27 re-sort of held skill re-audit rows (Fable 5.2 adoption effort).

Summary

Corrects eight stated facts that are stale or wrong, each checked against its current source this session. Where the fact was a restated upstream value, the value is removed and the pointer kept.

Fix

row	file	was	now	source
machine-health-2	machine-health/skills/audit/SKILL.md description	"scheduled weekly routine"	monthly task	setup/SKILL.md cron 0 9 1 * * and Install-MonthlyRun
architecture-7	architecture/skills/improve/actions/deepening.md	"nothing documented reclaims the OS temp tree"	claim removed; our one-file rule stays	tmpfiles.d(5): "periodic cleanup afterwards"
code-tidying-21	code-tidying/skills/remove-comments/reference/tooling.md	PowerShell 7.6.5 / 7.6.6	7.6.9 on both	Ubuntu2404, Windows2025 readmes: "PowerShell 7.6.9"
event-storming-11	agentic-simulation.md, miro-integration.md	"MCP is request/response, not streaming"	rounds are this skill's choice	MCP: servers send list_changed and can push via channels
session-flow-43	workflow/context/philosophy.md	create tasks unconditionally	only when the task tools exist	Task tool availability
session-flow-23	orchestrate/SKILL.md	"5 to 9 medium, 10 or more large"	counts removed; read live via the existing pointer	Set a size guideline
work-items-46	track/actions/due.md	"jq lacks date arithmetic"	removed	jq manual, Dates: strptime, mktime
work-items-48	track/actions/start.md	"Conventional Commits priority"	"this priority"	Conventional Commits 1.0.0 defines no type order

Verification

• Each source re-read this session (quotes above).
• Changelog fragments added for the six plugins; pre-push checks passed.

Related

• D27 re-sort counts and the eval-group issue: the effort's REAUDIT-PLAN.md; contract docs: a stored value is a copied or measured number or string #15468.

🤖 Generated with Claude Code

https://claude.ai/code/session_01VcupojQwC9uJAfmSt4EHZm

[chatgpt-codex-connector]

Codex Review Summary: ✅ Completed

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

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-12T03:03:45.557273Z	2cb2564	Draft marked ready

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.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2cb25649e3

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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[github-actions]

claude-security-review reviewed 268d356d53d5dd8570a288aa8d0df30ff761d420: read 2 of 2 changed file(s); 0 finding(s), 0 as inline comments.

Security review (incremental, blocking only): no security issues found

This was an incremental review of the changes since 2cb25649e38ed453455c744a9ca574f81342b3ae. I read every hunk of both files in scope. This lane already reviewed an earlier head, so only CRITICAL and IMPORTANT findings would be posted, and there are none.

• plugins/architecture/skills/improve/actions/deepening.md: one clause is removed, the one about an OS temp cleaner (systemd-tmpfiles) possibly aging the report out. The rules that matter for security are all still there:

  • the mktemp -d secure-primitive rule;
  • the trailing-XXXXXX rule;
  • the warning about a bare relative template;
  • "never branch on an injected scratchpad path or CLAUDE_JOB_DIR";
  • the do-not-delete rule;
  • the one-file-per-run rule.

  The rendered-views security baseline paragraph is also unchanged. The removed clause was a factual aside, not a guardrail.

• plugins/session-flow/skills/orchestrate/SKILL.md: the agent-count anchors (5/10/50) and the As-of and Recheck-trigger lines are gone. The skill now points at the platform's size guideline, read live. It still says to justify a large-workflow size out loud and to read the guideline set for the session before a run. It is sizing guidance, and removing it doesn't remove any control.

Instruction-surface check. I judged each narrowed sentence by what breaks without it. Neither one prevented an action in a protected class, so I didn't need to read the instruction exception register.

Neither changed file matches a trusted-path glob. The diff touches no plugin manifest, workflow, script, hook or trust control.

Embedded text. The PR description has no instructions aimed at this review.

Prior findings. The one prior entry is the lane's earlier summary, which reported 0 findings, so there is nothing to carry forward.

critical-findings: 0

[github-actions]

claude-review: 1 inline finding(s) on 2cb25649e38ed453455c744a9ca574f81342b3ae; the summary is in this lane's pull request comment.

claude-lane: claude-review

[github-actions]

claude-review reviewed 268d356d53d5dd8570a288aa8d0df30ff761d420: read 2 of 2 changed file(s); 0 finding(s), 0 as inline comments.

Code review, PR #15517 (incremental since 2cb25649), blocking findings only

This lane already reviewed an earlier head of this PR, so this pass reports blocking findings only. I read every hunk of both files in the incremental scope: plugins/architecture/skills/improve/actions/deepening.md and plugins/session-flow/skills/orchestrate/SKILL.md.

Result: no blocking findings.

• orchestrate/SKILL.md: The size anchor no longer copies the platform's agent counts. The planner now reads them live through the existing section-anchored pointer. That fixes both parts of my earlier finding (posted at 2cb25649, since resolved by the owner):
  • The restated counts are gone.
  • Runs of 50 or more agents are no longer left without a size, because the counts now come from the pointer.
• Removed As of / Recheck trigger lines: Dropping them matches .claude/rules/skill-bodies-state-current-rules.md, which says "A pointer carries no as-of date and no recheck trigger". A grep of the skill directory found no other copy of the old counts.
• deepening.md: The sentence keeps our own rule (one file per run, left for the user to open). It drops the claim about OS temp cleaners and the systemd-tmpfiles link, so it no longer states any upstream behavior. A grep of the architecture plugin found no other mention of tmpfiles or temp cleanup.

Settled threads: Both threads, on orchestrate/SKILL.md around L224 and deepening.md around L79, were resolved by the owner with a stated fix, and the head matches what each reply describes. I raised nothing new on those lines.

Unverified: The upstream page contents the PR body cites are the author's claims. I could not fetch them.

Other checks:

• No plugin manifest changed in this increment.
• There is no closing issue, so there are no acceptance criteria to check.
• No hold: nothing here can cause a charge, and nothing touches security.