llms.txt Audit & Generator
Audit any site for a missing, stale or malformed llms.txt — spec compliance, freshness and every link checked — then generate a compliant llms.txt from the live pages.
Free
Works with the AI Agents you already use
llms.txt Audit & Generator
Example session with this skill installed
Audit romankurnovskii.com for llms.txt
- Read your context and instructions
- Compiled the llms txt audit
Summary
romankurnovskii.com already ships a valid, fresh llms.txt at the site root — 121 links, 3 of them broken. No regeneration needed; the fix is the three dead links.
| Check | Result |
|---|---|
| File found | yes — https://romankurnovskii.com/llms.txt, HTTP 200 |
| Spec compliance | valid — H1 title, blockquote description, sections |
| Freshness | fresh — Last-Modified Tue, 03 Nov 2026 08:42:43 GMT |
| Links | 121 total, 118 reachable, 3 broken |
| Generated | no — the file was good enough to keep |
What the agent reports back
Where it lives. The file is served from the site root, so agents that probe /llms.txt find it on the first try. No redirect, no HTML fallback.
Spec. It opens with an H1 and a blockquote description, then groups pages under sections. That is what the standard asks for, and it is what an LLM parser expects.
Freshness. Last modified 03 Nov 2026, inside the 30-day window, and the linked pages still match the descriptions.
Link health. Three links out of 121 no longer resolve. They are listed individually in the report, so you can fix or drop each one instead of re-reading the whole file.
Raw report (trimmed)
{
"url": "https://romankurnovskii.com",
"llms_txt": {
"found": true,
"location": "https://romankurnovskii.com/llms.txt",
"status": 200,
"last_modified": "Tue, 03 Nov 2026 08:42:43 GMT",
"valid": true,
"fresh": true,
"validation": {
"has_h1": true,
"has_description": true,
"has_sections": true,
"total_links": 121,
"links_reachable": 118,
"links_broken": 3
}
},
"generated": false
}
What you do next
Open audit_report.json from the run, replace or remove the three broken entries, commit. If the file had been missing or stale, the same run would have written a fresh llms.txt alongside the report — this site did not need one.
Run against romankurnovskii.com on 2026-11-08.
Connects securely to your tools. The creator never sees your data.
About this skill
Your site can be perfectly indexed by Google and still be invisible to the agents that read it. llms.txt is the file that fixes that: a markdown map at the site root that tells an LLM which pages matter and what they contain.
This skill audits a site against the emerging llms.txt standard and fixes what it finds.
What it checks
- whether
/llms.txtexists at all, and where it actually lives - spec compliance: an H1 title, a blockquote description, section structure
- freshness: the file's Last-Modified date and whether the content still matches
- link health: every link in the file is fetched and the broken ones are listed
What it generates
If the file is missing or stale, the skill crawls the site's own pages, extracts the relevant ones, and writes a compliant llms.txt — a curated map rather than a full sitemap dump, with each entry described in one line.
How it runs
The skill ships a Python script (scripts/audit_llms_txt.py) that returns a single JSON report: what it found, what it validated, what it generated. The agent reads that report and explains the result — you do not parse the JSON yourself.
Where it stops
It does not host, deploy or serve the file: it gives you the llms.txt to commit. It audits publicly reachable pages; pages rendered only by client-side JavaScript may not be crawled. Generated maps are heuristic — read them before shipping.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.
Requires Python 3.10+. Target sites must be publicly accessible or credentials provided. Skips non-HTML assets like PDFs and ZIPs. Max crawl limit defaults to 50 pages.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 3 days ago
- Free with an Agensi account
- Software
Python
Git