docs(evi): fix four wording issues in the content-pass skill

[evlogai]

Applies the four clarify findings the 2026-10-06 repo health sweep (revision 88d2c9f2) confirmed on apps/evi/agent/skills/content-pass/SKILL.md, per EVL-506. Wording only: no procedure, bound, or description changes.

The four fixes

1. Stale corpus count (line 8). The scanner's own count (PR #783: 155 files) had outrun the hardcoded ~120.

-One pass, one group, one pull request. The corpus is ~120 files: the docs tree and the landing, the four package READMEs, the internal and published skills, and the three AGENTS.md files.
+One pass, one group, one pull request. The corpus is roughly 150 files: the docs tree and the landing, the four package READMEs, the internal and published skills, and the three AGENTS.md files. When the scanner's own count differs, the scanner wins.

2. No slot for the checks record in the PR-body template. Step 6 requires "Record which checks ran on which revision" but the template had no place for it, so the 2026-10-06 pass improvised a ### Checks section (PR #783). Added to the template, after ### Reported, not changed:

### Checks

- [what ran, on which revision, result]. One line each; omit only when the diff is markdown-only prose.

**3. blocked carried two senses in adjacent sentences (line 66): never reviewed, versus reviewed and refused.

-A `blocked` verdict caused by a failed snapshot load must be recaptured and reviewed before rewriting. A verified page blocked by critical findings is eligible for rewriting those findings.
+A `blocked` verdict from a failed snapshot load means the page was never reviewed: recapture it and send it back to review before rewriting. A verdict blocked on critical findings means the page was reviewed and refused: rewriting those findings is allowed.

4. Ready-versus-CI wording implied a draft flow the skill forbids (line 90). Step 7 says to open a normal, ready PR and report the blocker instead of opening a draft, so there is no draft to mark ready.

-... do not weaken thresholds or add filler to satisfy the score. Required CI failures still prevent marking the PR ready.
+... do not weaken thresholds or add filler to satisfy the score. A required CI failure blocks the PR: fix it or report the blocker.

Not changed

• Line 32's "a corpus of 120 files with 47 eligible pages" is a worked example, not a size claim, so it stays. The new "the scanner wins" sentence now covers the discrepancy it could otherwise read against.

Checks

• node scripts/content-lint/index.mjs apps/evi/agent/skills/content-pass/SKILL.md on main at 00862de after editing: exit 0, score 95. One finding (U-15, drain/sink at line 165) is pre-existing on an unrelated passage and a false positive against the file's own example.
• pnpm run lint on the branch at a8d1235: exit 0 (6 cached, 25 total).

No changeset: the diff is confined to apps/evi. Typecheck and test are unaffected by a prose-only change to one skill file and were not run.

ELI5

The written procedure the agent follows for daily docs cleanups had drifted from reality in four small ways: a file count that was out of date, a report format missing a section the same procedure requires, one word doing two jobs in back-to-back sentences, and phrasing that described a draft step the procedure elsewhere says not to take. This PR fixes all four sentences in that one file, changing nothing about how the pass actually runs.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
evi	Building	Preview	Oct 6, 2026 9:16pm UTC

4 Skipped Deployments

Project	Deployment	Actions	Updated
evlog-docs	Skipped	v0	Oct 6, 2026 9:16pm UTC
evlog-render-lab	Skipped		Oct 6, 2026 9:16pm UTC
evlog-telemetry	Skipped		Oct 6, 2026 9:16pm UTC
just-use-evlog	Skipped		Oct 6, 2026 9:16pm UTC

[coderabbitai]

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

• Configuration used: defaults
• Review profile: CHILL
• Plan: Advanced
• Run ID: 3a1e41bb-ab15-4a5e-8ed8-a77912ddae3d

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

• Autopilot · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

---

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.}

[github-actions]

Thank you for following the naming conventions! 🙏