API

Validate

Lint a file for structure and agent-readiness. Validate a served URL or raw content — handy in CI to check a file before you publish it.

Prefer a UI? The same checks run in the free web llms.txt validator — no API key needed.

text

POST /api/v1/validate

Request body

Provide either url or content:

Field	Type	Description
`url`	string	Fetch and lint the file served at this site.
`content`	string	Lint raw text directly (max 500 KB). No network fetch.
`checkLinks`	boolean	Sample links for liveness. Defaults to `true`.

bash

# Validate a live file
curl https://llms-txt.io/api/v1/validate \
  -X POST -H "x-api-key: $LLMSTXT_KEY" -H "content-type: application/json" \
  -d '{ "url": "https://example.com" }'

# Validate content from a build step before publishing
curl https://llms-txt.io/api/v1/validate \
  -X POST -H "x-api-key: $LLMSTXT_KEY" -H "content-type: application/json" \
  -d '{ "content": "# Example\n\n> Summary…", "checkLinks": false }'

Response

json

{
  "mode": "url",
  "url": "https://example.com",
  "targetUrl": "https://example.com/llms.txt",
  "score": 92,
  "readinessScore": 80,
  "status": "ok",
  "health": "live",
  "checks": [
    { "id": "has-title", "tier": "structure", "pass": true, "label": "Has an H1 title" }
  ],
  "usesInstructionsPattern": true,
  "hasLlmsFull": true,
  "contentLength": 1843,
  "linkStats": { "checked": 12, "ok": 12, "dead": 0 }
}

Field	Description
`score`	Structure/compliance score (0–100).
`readinessScore`	Agent-readiness score (0–100) — the higher-value signal.
`checks`	Per-check results, each tagged `tier`: `"structure"` or `"readiness"`.
`health`	`"live"`, `"degraded"`, or `"dead"` (URL mode only).
`usesInstructionsPattern`	Whether the file includes an agent-instructions block.
`hasLlmsFull`	Whether a companion `llms-full.txt` was found.
`linkStats`	Sampled link-liveness results (when `checkLinks` is on).

In CI, validate the content you just generated with checkLinks: false for a fast, network-light gate, then fail the build if readinessScore drops below your threshold.

Validation is cheap, so quotas are generous — see Rate limits & errors.