Jev.aitools.fyi
Agent and API surface
This page describes how software agents and integrators can read Jev Directory without scraping HTML. It is not the TypeSafe Jev inference API.
Human vs agent entry
The homepage hero and site header include a Human | Agent control. Human is the default visual experience. Choosing Agent opens /for-agents, a crawl-oriented page with the same discovery links as this doc. No JavaScript is required: the hero also exposes a plain I'm an agent link.
Markdown negotiation: send Accept: text/markdown to /for-agents, or add ?view=agent on / for the longer agent map. HTML at / is unchanged for typical browser Accept headers.
Start here
Canonical site URL: https://jev.aitools.fyi. Human readers use the same host; agents can negotiate Markdown on the homepage and fetch machine-readable indexes below.
- /openapi.json OpenAPI 3.1 description of discovery endpoints
- /llms.txt Plain-language site map for LLM crawlers
- /sitemap.xml Indexable HTML routes
- /search-index.json Listing search index used by the Explore UI
- /.well-known/jev-directory.json Publisher manifest with counts and URL patterns
Content negotiation
Send Accept: text/markdown to / for a Markdown overview. HTML remains the default for browsers. Responses include Vary: Accept.
Unknown paths return HTTP 404. With Markdown Accept, 404 bodies include links back to home, llms.txt, and the sitemap.
JSON errors under /api/*
This directory does not expose a write API on /api/*. Unknown routes return JSON with shape { "error": { "code", "message", "hint" } }. See OpenAPI for examples.
TypeSafe inference
To call System One (Jev) in your app, use TypeSafe credentials and official API docs. This site only links to community and official listings.