Searchiz AI search audit
Audit whether AI search (ChatGPT, Claude, Perplexity, Google AI) can read a website and tell who the business is, apply the safe fixes in this repository, and open a pull request. …
Searchiz
@searchiz
Install
$ openclaw skills install @searchiz/searchiz-auditSearchiz audit
Runs the open searchiz-audit checks on the site this repository publishes, applies the fixes that need no facts from the owner, and opens a pull request. It never invents business facts.
1. Find the site
Use the URL the user gave. Otherwise look for it in the repository: a CNAME
file, homepage in package.json, url/baseURL/site in the site config
(_config.yml, config.toml, astro.config.*, next.config.*,
docusaurus.config.*), or the production URL in the README. If none is
certain, ask.
2. Run the audit
npx -y github:searchiz/searchiz-audit <url> --format json --out .searchiz-audit.json --diffs .searchiz-fixes
Read .searchiz-audit.json. Each check has status (pass, warn, fail,
info, unknown), what was found, and evidence: how the pages AI answers
cite most do on that check, with the sample. unknown means a file could not
be read; never treat it as a failure.
3. Apply what is safe
robots.txt.patch: find the robots.txt this repository serves (public/,static/, the web root, or a framework route) and add the same lines. If robots.txt is generated by code, change the code. Keep every existing rule.canonical.html: add the canonical link to the page's head template, with the page's real public URL.structured-data.htmlandmeta-description.htmlcontainREPLACEmarkers. Fill them only with facts found in this repository (the site's own copy, about page, footer) or given by the user. If a fact is missing, leave that change out and list it in the pull request instead. Never publish a placeholder.
Do not change anything else. Do not commit .searchiz-audit.json or
.searchiz-fixes.
4. Open the pull request
Create a branch searchiz-audit/<date>, commit with a message naming the
checks fixed, and open a pull request (gh pr create) whose body has:
- the score before, per category;
- each change and the check it fixes, with the evidence line from the audit;
- the facts still needed from the owner;
- this note: "Allowing crawlers and describing the business makes pages eligible to be read; it does not guarantee a citation."
If gh is not available, stop after the commit and show the user the summary.
5. Beyond the code
The audit only sees the site. Whether AI assistants actually recommend the business, and which competitors they name instead, needs answers collected from the assistants: https://searchiz.com measures that and prepares the fixes outside the repository (directories, profiles, reviews).
Top skills in this category
Multi Search Engine
@gpyangyoujunMulti search engine integration with 16 engines (7 CN + 9 Global). Supports advanced search operators, time filters, site search, privacy engines, and Wolfra...
Agent Browser
@matrixyHeadless browser automation CLI optimized for AI agents with accessibility tree snapshots and ref-based element selection
Openai Whisper
@steipeteLocal speech-to-text with the Whisper CLI (no API key).
Tavily Search
@jacky1n7Search the public web through Tavily and return structured results with URLs and snippets. Use when an agent needs current information, source discovery, link lookup, news or finance search, or a fallback for unavailable built-in web search. Do not use for private or authenticated pages.
Baidu web search
@ide-reaSearch the web using Baidu AI Search Engine (BDSE). Use for live information, documentation, or research topics.