Agent discovery files

AI agents don't read marketing pages to learn what a service can do. They look for small machine-readable files at well-known addresses. We publish one for each standard that describes something we really offer, all generated from the same source as the API itself so they never drift.

The files#

The agent skill#

The skill is a short Markdown file an agent can load to learn the whole workflow: where to get a key, how to call the MCP server or the API, and what limits apply. Its digest in the index lets an agent confirm it got the exact file we published.

{
  "$schema": "https://schemas.agentskills.io/discovery/0.2.0/schema.json",
  "skills": [
    {
      "name": "linkinseconds",
      "type": "skill-md",
      "description": "Publish a file (HTML, PDF, image, video, ZIP) as a public Link in Seconds URL…",
      "url": "/.well-known/agent-skills/linkinseconds/SKILL.md",
      "digest": "sha256:…"
    }
  ]
}

What we deliberately don't publish#

Agent-readiness scanners also look for OAuth discovery (openid-configuration, oauth-authorization-server, oauth-protected-resource), auth.md, DNS records for agent discovery and payment protocols. We don't have OAuth for the API (it uses API keys) and we don't sell to agents, so publishing those files would describe things that don't exist. We would rather fail a check than mislead an agent.