agentleFS
XPOZpublicXPOZpublic/geo-seo-agent/.claude/skills/content-rework/SKILL.md

content-rework

Skill0 starsChanged 3 months agoMITView on GitHub

Surgically improve an existing published page per an approved rework brief. Fix what the diagnosis names while preserving the page's existing rankings. Do no harm.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Never invents reworks. One brief per run.

    line 13
  2. Very specific:

    Writing principles: follow strategy/geo-writing-principles.md for openings (never lead with your brand), brand placement, extractability, …

    line 21
  3. Very specific:

    Read brief + diagnosis; validate approved.

    line 25

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Content Rework

Improves an existing published page per an approved brief (action: rework). The prime directive is do no harm: the page has equity (queries it already ranks for, an indexed URL with age), and a careless rewrite loses more than it gains.

Boundaries

  • Requires an approved row (in content/index.md for site pages, content/docs/index.md for docs pages) with action: rework and a brief.md containing the live URL and a diagnosis: what specifically is failing (gap analysis's page-audit findings, e.g. no answer-first opening, no data points, stale facts, weak title). No diagnosis → stop; a rework without a diagnosis is a rewrite in disguise.
  • Never invents reworks. One brief per run. Never publishes; the human (or the company's publishing workflow, with its human gate) applies the changes to the live site.

Do-no-harm rules

  • Never change the slug/URL. A rename is a separate, human-decided operation (redirects involved) and out of scope.
  • Read before cutting. Before editing, capture the page's current state: fetch the live page, and where Search Console is connected pull its current winning queries (clicks/impressions/position). Write a before-state note into the item's folder (before-state.md). Sections and phrasings serving winning queries are load-bearing: improve around them, never delete them. Without Search Console the before-state is the fetched page itself plus whatever ranking evidence the snapshot holds; say so in the note.
  • Targeted surgery, not rewrite. Fix exactly what the diagnosis names; keep everything else intact. If the diagnosis turns out wrong, or the page genuinely needs a ground-up rewrite, stop and report: that's a new decision for the human, not a bigger edit.
  • Freshness honestly. The modified date updates to today; the original publish date never changes. Verify every fact you update with fresh web research. Your own product facts come from strategy/company-profile.md.
  • Writing principles: follow strategy/geo-writing-principles.md for openings (never lead with your brand), brand placement, extractability, and competitor tone (factual-neutral, no loaded language).

Process

  1. Read brief + diagnosis; validate approved.
  2. Fetch and record the before-state into the item folder's before-state.md.
  3. Research fresh for the facts being updated. Check memory/competitors/ first for source-verified starting context; re-verify anything older than ~4 weeks. Use WebSearch/WebFetch, never the ask scripts.
  4. Produce the reworked page as reworked.md in the item's folder, plus a short change summary (what changed and why, mapped to the diagnosis) the human can review before applying.
  5. Hand off per the company profile's publishing workflow, or deliver the reworked draft and change summary to the human.
  6. Record: index row updated, journal entry, state.md if changed.

Success definition

Different from generation: the rework worked only if its target prompts moved and the before-state's winning queries didn't drop. The before-state note is what makes that regression check possible; measurement and retro weigh it weeks later, and this skill's job is capturing the baseline.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.