Read this site the easy way.

Markdown for agents

Send Accept: text/markdown to any page URL and you get a Markdown rendering of that page instead of HTML, with the title and description as front matter. The HTML page stays the default for browsers, and the URL does not change.

curl -H "Accept: text/markdown" https://twimco.com/journal/

Responses carry x-markdown-tokens and x-original-tokens so you can see what the conversion saved before spending context on it.

Discovery

The API catalog at /.well-known/api-catalog follows RFC 9727 and points to this page, the OpenAPI description and the status endpoint. Crawl policy lives in robots.txt, which also declares Content Signals, and every published URL is listed in sitemap.xml.

Content API

Published content is readable without credentials at https://twimco.com/api. Unauthenticated requests see published documents only. Drafts are never served.

• GET https://twimco.com/api/posts — Editorial notes published at /journal/.
• GET https://twimco.com/api/pages — Standalone site pages and their editable copy.
• GET https://twimco.com/api/services — Service pages published under /services/.
• GET https://twimco.com/api/case-studies — Project write-ups published under /work/.
• GET https://twimco.com/api/projects — Short project summaries shown on /work/.
• GET https://twimco.com/api/media — Images and their alt text, captions and credits.
• GET https://twimco.com/api/globals/{name} — site-wide content: homepage, navigation, footer.

Lists accept limit, page, sort, depth and where. To fetch one field note by slug:

curl "https://twimco.com/api/posts?where[slug][equals]=an-agent-needs-a-job"

Article bodies are stored as structured rich text. If you want prose rather than nodes, request the page itself as Markdown.

Sending an introduction

POST https://twimco.com/api/contact accepts the same introduction the contact page sends. Generate a UUID for id and reuse it if you retry, so a repeated send returns the original reference instead of creating a second inquiry. The response carries a reference to quote when following up.

If you are a person rather than a program, the contact page is a better experience than the endpoint.

Authentication

There isn’t any. Published content and the introduction endpoint both work without credentials, there is no agent registration endpoint, and this site runs no OAuth authorization server. auth.md says so in machine-readable form, so an agent can confirm it in one request. Anything beyond the public surface is arranged with a person through the contact page.

Limits and etiquette

• Introductions are rate limited per client address.
• Keep depth low and limit modest; deep, unbounded queries are slow for everyone.
• GET https://twimco.com/api/health/ reports whether content is being served.
• Identify your agent in the User-Agent header so we can tell you apart from a scraper.