No agent-relevant Link headers

What it detects

HTTP responses carry no Link header advertising an agent-relevant relation (api-catalog, service-desc, service-doc).

Why it matters

A Link header is a machine-discoverable pointer to a site's programmatic interface that costs nothing to check — its absence means an agent has no cheap way to discover whether an API exists before falling back to scraping.

How to fix

- Fetch the homepage with `curl -I` and check the response for a `Link:` header carrying an agent-relevant relation type (`api-catalog`, `service-desc`, or `service-doc`). - Add a `Link` response header pointing agents at your machine interface, e.g. `Link: </.well-known/api-catalog>; rel="api-catalog"` (RFC 9727) — set it in your web server / CDN config (nginx `add_header`, a middleware, or an edge rule) so it rides on every HTML response, or at minimum the homepage. - If you already publish an OpenAPI spec, `Link: </openapi.json>; rel="service-desc"` is the lowest-effort option; publish the linked resource (`/.well-known/api-catalog` as a `linkset+json` document, or your existing spec) so the pointer resolves. - Verify: `curl -sI <homepage> | grep -i '^link:'` and confirm at least one of the three rels appears and its target URL returns HTTP 200.

References

• RFC 9309 — Robots Exclusion Protocol
• Sitemaps protocol