agentleFS
ucsandmanucsandman/claude-skills/ai-search-optimization/SKILL.md

ai-search-optimization

Skill5 starsChanged 2 months agoMITView on GitHub

Improve retrieval, citations, and accurate representation in AI search answers.

2 other projects use a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write headings as the questions people actually ask.

    line 27
  2. Distinctive:

    Keep facts consistent across your site.

    line 29
  3. Distinctive:

    Leave an inaccurate representation uncorrected because it is not on your site.

    line 52

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

AI search optimization

Classical SEO optimizes to be clicked. This optimizes to be quoted — often with no click at all. That changes what a good page looks like.

What gets cited

  • Self-contained passages. A retrieved chunk arrives without the surrounding page. Each section must make sense alone, with its subject named rather than pronominalized.
  • Direct answers near the question. Bury the answer under three paragraphs of context and the passage retrieved will be the context.
  • Specific, checkable facts — numbers, dates, named methods, stated conditions. Vague claims are neither retrievable nor quotable.
  • Attributable expertise. Named authors, stated credentials, dated content, and cited sources. Anonymous undated content is weakly weighted.
  • Structure that survives extraction — real headings, real lists, real tables. Layout implied by styling disappears.

Practical moves

  • Answer the question in the first sentence under each heading, then elaborate.
  • Write headings as the questions people actually ask.
  • Define your own terms on your own pages, so the model's definition traces to you.
  • Keep facts consistent across your site. Contradictions get resolved against you.
  • Maintain the boring canonical pages — pricing, comparisons, specifications, FAQ. These are heavily retrieved and usually neglected.

Being represented accurately

Assistants assemble an answer about you from whatever is available, weighted toward third-party and structured sources. Where those are thin or stale, the answer will be wrong.

Audit periodically: ask several assistants what your company does, who it is for, what it costs, and how it compares. Note the errors and trace them to a source. The fix is almost always publishing or correcting the source, not the assistant.

Measuring

Click-through will fall on informational queries even as influence rises. Track citation and mention frequency, and downstream branded search and direct traffic, rather than judging this program on organic sessions — that metric will say you are losing while you are winning.

Never

  • Optimize for one assistant's current behavior. Retrieval and citation rules change without notice and without a changelog.
  • Assume being crawlable means being citable. Models cite sources that answer a question cleanly, not sources that merely exist.
  • Leave an inaccurate representation uncorrected because it is not on your site. The claim propagates whether or not you own the page it came from.

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.