mod_agent_discovery

Advertise bounded anonymous reads and existing resource export through /llms.txt, /.well-known/api-catalog, /openapi.json and /api-help.

Enable this module on public publishing sites. Templates using the standard {% all include "_html_head.tpl" %} hook receive discovery links automatically. Discovery stays in HTTP headers, the HTML head and dedicated documents; no links are added to human-facing page bodies or footers. Resource response headers use the same links and preserve existing Link values.

Discovery metadata and resource visibility checks always use an anonymous context. Existing model APIs retain their own ACL and authentication behavior. This module does not change permissions, robots rules or API indexing headers. Additional JSON-LD and Turtle support awaits the SPARQL implementation merge.

The description configuration is optional public text describing the site. It is not a license grant. Site templates can override agent_api_help.tpl.

Discovery model: m.agent_discovery.info and m.agent_discovery.links[id]. Content model: m.agent_content provides bounded search, resource, related and schema reads. Markdown uses category-aware _agent_content.tpl composition and _agent_block_<type>.tpl renderers. Survey blocks are always excluded. Content routes support GET/HEAD and require ETag revalidation after anonymous permission/publication checks. See the agent discovery guide for request limits.