The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Aigentify listing page.
aigentify audits a product directory or live URL for agent-native readiness. It also creates the public artifacts that agents can use.
Use the library in JavaScript or TypeScript:
Use the CLI:
Use the MCP server with an MCP client:
The MCP server exposes audit with { target } and gen with { artifact, params }.
aigentify audit [target] checks a directory or URL. The default target is ..aigentify init writes starter AGENTS.md and server.json files. Existing files stay unchanged.aigentify gen <artifact> prints an artifact or writes it with --out.aigentify mcp starts the stdio MCP server.Artifacts are agents-md, server-json, offer, and agents-route.
audit scores these signals. The core signals work for a directory or a URL; the
web signals are only meaningful for a served URL and are marked na otherwise.
Core: agents-md, agents-md-public-safe, server-json, mcp, offer-jsonld,
honest-offer, llms-txt, sitemap.
Web (URL targets): well-known-agent, openapi, agents-page, soft-404
(unknown paths must return a real 404, not a 200 app shell), markdown-negotiation
(serve text/markdown on Accept: text/markdown with Vary: Accept),
content-without-js (an <h1> and 500+ characters without JS), org-schema
(Organization JSON-LD with contactPoint and address), crawler-reachable
(agent User-Agents can reach the homepage), and rate-limit-headers (standard
RateLimit-* headers), plus auth-md, api-catalog (RFC 9727),
agent-card-a2a (A2A), link-headers (RFC 8288), markdown-alt, plus external-discovery checks that query third-party registries — ai-plugin, wikidata (P856 official-website match), npm-package, and mcp-registry.
Add an optional aigentify.config.ts, aigentify.config.js, or aigentify.config.mjs file. Export defineConfig(...) data as the default export or as aigentifyConfig.
MIT