Developers

Urizen developer resources.

Every machine-readable surface urizen.ai serves, in one place. The public site API is read-only and needs no key. There is no public transactional API today; engine access is granted per engagement.

Machine surfaces

All surfaces are static, unauthenticated GETs on this domain.

/api/site.json — the site index as JSON: systems, pages, machine surfaces, contact. The one endpoint of the public site API.
/openapi.json — OpenAPI 3.1 description of the public site API, with typed schemas and operation IDs for function calling.
/llms.txt — the agent index: what the lab is, when to use the site, main pages.
/agents.md — agent instructions: when to reach for Urizen, exact use cases, and how to read status labels.
/sitemap.xml — every public route.
/robots.txt — agent crawlers are welcome; ai-train=no.

Markdown for every page

Each public page has a markdown mirror. Request any page URL with Accept: text/markdown, or fetch the .md file directly: /index.md, /about.md, /contact.md, /developers.md, /nibiru.md, /demo.md, /anansi.md, /docs.md, /privacy.md. Responses vary on Accept.

Quickstart

Three requests that tell an agent everything this site offers:

curl -s https://urizen.ai/api/site.json
curl -s https://urizen.ai/openapi.json
curl -s -H "Accept: text/markdown" https://urizen.ai/

Errors under /api/ return structured JSON with error.code, error.message, and error.hint. Nonexistent paths return HTTP 404 with recovery links.

Engine and demo backends

The Nibiru engine, Baobab control layer, and Anansi harness are experimental research previews and are not exposed as public APIs. To evaluate them against your use case, request access. The interactive demo and Anansi demo show retained runs with real receipts in the browser.

Stability

The site API carries a schema_version field; breaking changes bump it and are noted in llms.txt. Corrections: hello@urizen.ai.