refresh
Skill1 starsChanged 6 months agoApache-2.0View on GitHub
Upgrade an existing blog post — re-run SERP analysis against the current top 3, refresh stats from `stats.md`, fix any Tier 1 on-page SEO gaps, update internal links to recent content, refresh the `dateModified`. Preserves the original URL and primary keyword (no cannibalization). Use when the user types `/refresh` or asks to update an existing post.
13 other projects use a version of this.
Adopted by
Most recently taken up by
- adelrodriguez/skillstheir own version, first seen Nov 15, 2026
- ai-search-guru/getcito-worlds-first-open-source-aio-aeo-or-geo-tooltheir own version, first seen Nov 8, 2026
- kyle-chalmers/ticketwrighttheir own version, first seen Nov 8, 2026
- ansvisor/ansvisortheir own version, first seen Nov 1, 2026
- RightBlogger/bloggingskillstheir own version, first seen Nov 1, 2026
- zoo3323/usage-statuslinetheir own version, first seen Oct 15, 2026
Used with
Projects with it also use setup.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
VercelSkillsetup
Set up Vercel CLI and project configuration. Use when the user says "set up Vercel", "configure Vercel", "link to Vercel", or "vercel init".
Also in 6 projects that use this
Other versions13 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 13
- ai-search-guru/getcito-worlds-first-open-source-aio-aeo-or-geo-tool★ 432same name
- ansvisor/ansvisor★ 241same name
- iliasedelkin/jobissimo★ 8same name
- fullymiddleaged/Clawness★ 5same name
- kyle-chalmers/ticketwright★ 4same name
- RightBlogger/bloggingskills★ 4same name
- prPMDev/my-portfolio-kit★ 2same name
- adelrodriguez/skills★ 1same name
- fabio-baumgartner/unternehmer-skills★ 1same name
- NikolajSaudella/cue★ 1same name
3 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository20 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- alternativeSkill
- auditSkill
- blogSkill
- broken-backlinksSkill
- case-studySkill
- clusterSkill
- comparisonSkill
- competitor-backlinksSkill
- glossarySkill
- haroSkill
- integrationsSkill
- lovable-deploySkill
- pricingSkill
- programmatic-batchSkill
- semrush-baselineSkill
- serp-featuresSkill
- serviceSkill
- triageSkill
- wireframeSkill
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 3 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 11 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 24
If the invocation contains --cron, this skill runs in cron mode (no user available).
- Very specific: line 30
Defaults: when a choice would normally be asked, default to site-info.md / goals.md values.
- Very specific: line 32
Audit log: ALWAYS write one line to the audit log on exit — success (shipped), escalation (escalated), idempotent skip (idempotent-skip), …
- Very specific: line 77
Read site-info.md Languages section.
- Distinctive: line 100
If the post predates the AI-search rules (most posts shipped before 2026-05-17), assess and apply:
- Distinctive: line 103
Remove "as mentioned above" / "see below" references — restate context inline (self-contained facts)
- Distinctive: line 104
Add inline source links to any unsourced statistics.
“Common” is measured: kept by at least 6 in 10 of the 11 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
/refresh — Existing-post refresher
Upgrades an existing published blog post in place. Often higher-leverage than writing a new post — Google rewards freshness on existing-ranking content.
Inputs
Required:
- Site — must match a folder under
sites/[name]/. - Post URL — the live URL of the post to refresh.
Optional:
- Refresh scope —
light(stats + internal links +dateModifiedonly),medium(light + Tier 1 fixes + image upgrades),heavy(medium + section rewrites + new SERP analysis). Default:medium. - Image source — overrides default (only if scope includes image upgrades).
Workflow
Step 0 — Cron-mode detection (if invoked with --cron)
If the invocation contains --cron, this skill runs in cron mode (no user available). Required behavior:
- Idempotency check: see this skill's "Routine versioning + idempotency contract" section. If today's output already exists, exit cleanly with
exit: "idempotent-skip". Write one line to the audit log viascripts/lib/audit-log.mjsappendRun({ exit: "idempotent-skip", ... }). - Escalation contract: any decision that would normally prompt the user (missing required file, voice anti-AI failure, schema validation failure, keyword cannibalization, etc.) MUST be escalated by writing an item to
sites/{site}/_inbox/viascripts/lib/cron-mode.mjswriteInboxItem(...). After writing, exit cleanly withexit: "escalated". Do NOT useAskUserQuestionin cron mode. - Defaults: when a choice would normally be asked, default to
site-info.md/goals.mdvalues. If both are silent on the required choice, escalate per (2). - Audit log: ALWAYS write one line to the audit log on exit — success (
shipped), escalation (escalated), idempotent skip (idempotent-skip), or failure (failed). - Backoff: at the start of every cron-mode run, call
checkBackoff({ routine })fromscripts/lib/audit-log.mjs. If true, the routine has hit the 3-strike threshold — write_inbox/routine-disabled-{name}.md, runscripts/sync-schedules.mjs --pause-routine={name}, and exit.
Step 1 — Read context (Rule 1)
Same as /blog Step 1.
Step 2 — Fetch the existing post
- For
repo-commitsites: read the source file from the site's repo (markdown/MDX). - For other publishing methods: scrape the live URL → extract title, meta, body, schema, image references.
Identify:
- Current primary keyword (from URL slug + title)
- Current cluster (inferred from H2s + body)
- Word count, image count, internal/external links, schema present
- Last published date + last modified date (if visible)
Step 3 — Run audit on the post
Score current state:
- Tier 1 status (which of the 16 pass / fail)
- Schema status (validates? complete?)
- Voice status (anti-AI check on existing copy)
- Stats freshness (any numbers that look stale or rounded against
stats.md?) - Internal links (do they still point to live URLs? are there newer related posts to link to?)
- External links (still resolve? still authoritative?)
- Length vs. current SERP top-3 median (re-run SERP analysis fresh)
Present scorecard to user.
Step 4 — Plan refresh + approval (Rule 2)
Based on scope + audit:
light→ list of stat updates + internal link updates +dateModifiedbump.medium→ light + specific Tier 1 fixes + image upgrades for any failing 6.4 (width/height) or 6.3 (WebP under 200KB).heavy→ medium + rewritten sections (specifically: opening if it doesn't pass voice, FAQ if missing or schema-mismatched, any H2 sections that the new SERP analysis suggests are missing or off-target).
Present plan. Wait for approval.
For multilingual sites, the plan also lists which language counterparts are being refreshed. Default: refresh ALL language versions of the page in lockstep — translation drift is real and the toolkit prevents it.
Step 4.5 — Resolve language fan-out (multilingual sites only)
Read site-info.md Languages section.
Multilingual: false→ skip; continue to Step 5 as a single-language refresh.Multilingual: true:- For the post being refreshed, locate every language counterpart (use the URL pattern table from
site-info.md). If a counterpart is missing, FLAG it as a coverage gap and refuse to refresh until either (a) the missing version is created, or (b) user explicitly confirms partial-coverage opt-in. - Apply the SAME refresh scope to every language version. If only the EN version had a stat-staleness issue, still bump the ES version's
dateModifiedand re-resolve its voice/anti-AI checks against the currentvoice.[lang].mdrules. - Drift detection: compare structural shape (H2 count, FAQ count, schema entities) between language versions. If they've drifted (one has 6 FAQs, the other has 4), surface as a refresh item.
- Hreflang verification: confirm reciprocal hreflang link cluster is still correct after the refresh. The slug is unchanged (per Rule preservation), so hreflang URLs should remain stable — but verify.
- For the post being refreshed, locate every language counterpart (use the URL pattern table from
The refresh produces N updated drafts (one per language), each in its own _drafts/[slug]/[lang]/ subfolder. Ship them together.
Step 5 — Apply refresh
Critical preservation rules:
- ✅ Same URL slug — never change.
- ✅ Same canonical — preserved.
- ✅ Same primary keyword — never swap (cannibalization risk).
- ✅
datePublishedunchanged;dateModifiedset to today. - ✅ Author byline preserved unless explicitly asked to change.
Apply changes per scope.
AI-search optimization retrofit (apply during refresh — references/voice.md Universal AI-search optimization rules):
- If the post predates the AI-search rules (most posts shipped before 2026-05-17), assess and apply:
- Convert 3-5 H2/H3s to explicit question form where natural (Q+A density)
- Split paragraphs longer than 3 sentences (citation-friendly chunking)
- Remove "as mentioned above" / "see below" references — restate context inline (self-contained facts)
- Add inline source links to any unsourced statistics; drop unsourceable numbers (verifiable claims)
- These are upgrade, not preservation — explicitly tag in the refresh plan (Step 4) so the user reviews before applying. Do NOT silently rewrite a post's voice; the original voice stays.
Step 6 — Validate (Rule 4)
Same as /blog Step 8, with one addition:
- ✅ No URL change — verify slug unchanged.
- ✅ No primary keyword swap — verify primary keyword unchanged.
Step 7 — Ship per publishing method
Same as /blog Step 9, but as an UPDATE to existing content:
repo-commit→ commit torefresh/[slug]branch, open PR (or update directly).cms-paste→ output is the updated body + meta + image replacements; user updates the existing CMS entry.lovable-prompt→ output is a Lovable-ready prompt to update the existing page.headless-api→ updates the existing CMS entry via API (preserves the entry ID, updates fields).
Step 8 — Update tracker
In used-keywords.md, don't append a new row — instead, find the existing row for that primary keyword and append a "Last refreshed: YYYY-MM-DD" note. If the row doesn't exist (e.g., the post pre-dates the toolkit), add it with both dates.
Step 9 — Report
Print:
- What changed (light / medium / heavy summary)
- Specific Tier 1 fixes applied
- New internal links added
- Stats refreshed (with old → new values)
- Lighthouse delta if checked
- Reminder: request indexing in GSC after the live update propagates.
Routine versioning + idempotency contract
This skill participates in cruise-control via --cron mode. When invoked by cron:
- Stamp
routine_versionin every output produced (thenotes.mdaudit entry header,_inbox/item frontmatter, refreshed-draft folder metadata, audit log line viascripts/lib/audit-log.mjs). Currentroutine_version: 1.0. - Idempotency: this skill MUST be safe to run twice on the same target URL on the same day without producing duplicate work. Implementation: check whether
notes.mdalready records a refresh of this URL today; if so, exit cleanly withexit: "idempotent-skip".
Reference: docs/specs/2026-05-16-agents-cruise-control-design.md § Operational hardening O3.
Refusal conditions
- Post URL doesn't resolve.
- Refresh would change the primary keyword (refer user to
/blogfor new content instead). - Refresh would change the URL slug (would create a redirect chain — user must explicitly opt in via "I accept the redirect" flag).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

