Claude Skill A Claude Skill that pulls all your published URLs and GSC data, and ranks them by refresh opportunity into a working queue.

Refresh Queue Builder Claude Skill

Refresh Queue Builder is a free Claude Skill that turns your blog folder and your Search Console data into a ranked refresh queue. It pulls every published URL, joins GSC data per URL, scores each by refresh opportunity, and returns a markdown checklist with the highest-opportunity post on top. The SEO Refresher works from that top row.

What does the Refresh Queue Builder skill do?

It decides which post gets refreshed next, and writes that decision down as a queue. The first sentence of the Skill file’s description draws the division of labor: the SEO Refresher is the engine, this Skill is the intake. Four steps:

  • Step 1, corpus. You hand it your blog folder and your GSC dataset. The Skill asks for both explicitly; a missing input gets requested, and placeholders are banned.
  • Step 2, score. Each URL is scored on striking-distance impressions, time since last update, position trend, and click trend. Anything odd gets surfaced as a flag, and the Skill waits for your confirmation before ranking.
  • Step 3, rank. Highest opportunity at the top. Posts already refreshed get marked done rather than silently dropped. The output structure is fixed: every required section present and non-empty.
  • Step 4, queue. A markdown checklist, one row per URL, sorted by score. The SEO Refresher picks from the top.

The confirmation pause in step 2 is where your judgement enters. A URL can score high on search signals and still be the wrong refresh, a page whose traffic never converts, say. Whether those clicks are worth chasing is a call for you and your analytics, made at the flag stage.

How does it score refresh opportunity?

Four signals per URL. Each catches a different candidate. Striking-distance impressions find pages Google already shows for queries they never win, the cheapest lift in SEO. Time since last update finds pages going stale untracked. Position trend and click trend catch decay while it is shallow, and a post sliding on both gets an explicit declining-trend flag, a behavior written into the Skill’s hard rules.

Scoring from your own Search Console data is what keeps the queue honest. The file bans invented data outright: no GSC dataset, no scores, and the Skill asks for the export instead of estimating. The queue ranks the opportunities your property verifiably has rather than the ones somebody remembers.

What does the ranked queue look like?

A checklist you can read top to bottom and act on top-first. Here is a three-row sample; every value is invented for the illustration, and the row shape is the Skill’s own: slug, score, top striking-distance query, expected lift, last updated.

## Refresh queue, built 2026-11-06

- [ ] /blog/content-audit/ | 87 | "content audit checklist" | lift: p12
      toward page 1 | updated 2026-02-14 | FLAG: clicks declining to almost none
- [ ] /blog/content-inventory/ | 74 | "content inventory template" |
      lift: p9 toward top 5 | updated 2026-04-02
- [x] /blog/content-calendar/ | refreshed 2026-10-12, inside 90 days

Trace the rows to the file’s rules and each one earns its place. The sort is descending by score. Row one carries the declining-trend flag the hard rules require. The third row shows the 90-day rule: a post refreshed within the last 90 days is skipped for scoring and marked done, so the queue never re-assigns work before the last pass could register. And the ranking caps at the top 50 URLs by default, which keeps a thousand-post corpus readable.

The top row is no accident. The SEO Refresher’s page refreshes that same invented post in its own worked example, which is the handoff working.

How does the queue feed the SEO Refresher?

Directly. The checklist this Skill returns is the queue file the SEO Refresher reads: that Skill never chooses a post for you, with one exception, a queue file with checkbox rows in your repo, where it takes the topmost unchecked row. Build the queue here, save it in the repo, and each Refresher run consumes the next row without anyone re-litigating priorities.

That split is deliberate. Choosing the refresh and performing it are different jobs with different failure modes, so the series keeps them in separate files. One run here sets the order for weeks of refresh work; each Refresher run stays small, reviewable, and one post wide.

What will it refuse to do?

The boundaries are written into the file as hard rules.

  • No invented data. Scores come from your corpus and your GSC dataset; a missing input gets a question, never a stand-in.
  • No re-queuing fresh work. URLs refreshed in the last 90 days are skipped and marked done, never re-scored into the ranking.
  • No unbounded dumps. Top 50 by default. More rows arrive only when you ask for them.
  • No invented statistics, citations, or extrapolated claims. The queue reports what the data shows, and the declining-trend flag is raised from real trend lines.
  • No output that fails its own self-check. The Skill verifies its rules before returning and refuses to hand back a queue that breaks them.
  • No refreshing. It stops at the queue. Running the refresh is the SEO Refresher’s job, and pushing this Skill past that boundary gets a refusal plus a pointer to the Skill in the series that owns the work.

These refusals make the queue shorter. Short is the feature. A queue you trust gets worked; a queue padded with guesses gets ignored by week two.

Who is this Skill for?

People whose refresh backlog outgrew their ability to eyeball it.

  • A head of content with 300 published posts and no ranked view of which ten matter this quarter.
  • An SEO owner whose refresh spreadsheet went stale.
  • A solo marketer running a blog from Claude Code who wants the choosing done by data before the week starts.
  • A team already running the SEO Refresher whose intake is still a judgement call made in a meeting.

How do you install the Refresh Queue Builder skill?

Three steps, specific to this Skill’s inputs.

  1. Save the SKILL.md file to ~/.claude/skills/refresh-queue-builder/SKILL.md on your machine.
  2. In Claude Code, invoke /refresh-queue-builder and point it at your blog folder and your Search Console dataset when it asks.
  3. Review the flags it surfaces, confirm, and save the returned checklist into your repo where the SEO Refresher can read it.

First Skill you have installed? The series-wide install guide on the skills hub covers the folder layout, invocation, and claude.ai.

What’s in the bundle?

Four pieces come with the download.

  • SKILL.md (raw): the complete Skill file, destined for ~/.claude/skills/refresh-queue-builder/.
  • One-pager PDF: a printable page with the four steps, the queue row format, and the install path.
  • Google Doc (gated): a Drive-resident copy plus install notes. Submit your email to access; we also tag you for the Claude Skills for content operations email sequence so you hear when the next Skill lands.
  • Card image: the hero, free to repost.

Why we built it

Refresh Queue Builder is Skill 10 of 10 in Relato’s Claude Skills for content operations series, which covers the lifecycle from brief through refresh. It is last by number and first by position in the refresh loop, because our own loop kept failing at the same spot: the refresh was systematized while the choice of what to refresh stayed guesswork. Whoever ran the loop picked a post they remembered, and the posts nobody remembered decayed quietly. So we wrote the intake down. The choosing became as repeatable as the refreshing.

It stops at the queue. Running the refresh is the SEO Refresher’s job.

Frequently asked questions

What is a Claude Skill?

A file that teaches Claude Code a repeatable job. The SKILL.md carries frontmatter naming the Skill and a body holding its steps and hard rules. Save it into your skills folder once, and from then on Claude Code loads it whenever you invoke it by name.

Do I need a paid Claude plan to use this?

Claude Code is the one requirement. The Skill itself is a plain file under an MIT license, with no account, server, or signup behind it.

Does it work without Google Search Console data?

No. The scores rest on impressions, positions, and click trends only your own GSC property holds, and a hard rule bans inventing data. Without the dataset, the Skill asks instead of estimating.

Why does it skip posts refreshed in the last 90 days?

Because a refresh needs time to register before it is worth re-scoring. The SEO Refresher’s method note treats 4 to 8 weeks as the window in which a position move shows up, so a post refreshed last month carries no readable signal yet. The rule keeps the queue from resurfacing work that has not had its chance.

Can it rank more than 50 URLs?

Yes, when you ask. The default is the top 50 so the queue stays reviewable in one sitting, and the rule lives in plain markdown inside the SKILL.md, so you can raise the cap permanently by editing the file.

Will more Skills follow?

The series runs to ten and this Skill completes it. The next Skill in the workflow is where your queue goes: the SEO Refresher, Skill 2, takes the top row and runs the refresh. And the email sequence behind the form above walks the full series one Skill at a time.

Ready to streamline your content operations?

Start your free trial and see the difference.