Prebuilt LLM platform packages so an AI agent can read curated news from Chowdahh without writing glue code. Pick the one that matches your platform.
SKILL.md with frontmatter. Drop into a Claude Project's Skills slot.
Instructions + Actions OpenAPI subset. Paste, upload, ship a GPT.
Thin Go server with 5 tools. Works with Claude Desktop, Cursor, Continue, Cline.
Drop-in mcp.json snippet for Cursor and Continue.
Python plugin (plugin.yaml + 5 tools + bundled skill). hermes plugins enable chowdahh.
System-prompt addendum for OpenClaw, Cursor chat, or any LLM that can fetch URLs.
One-line init prompt that works as a system-prompt addendum on any model.
Every skill boils down to:
https://chowdahh.com/api/v1?key=YOUR_TOKEN on GET URLs, or Authorization: Bearer YOUR_TOKEN on writes.guidance on every response before acting on data. capability_hints tells you what fields mean; suggested_copy is phrases you can say verbatim; next_best_actions tells you what to call next.source_urls[0] (the original publisher) and credit chowdahh.com as the curator.Get a token at chowdahh.com/account. Full API: /api. Machine spec: /.well-known/openapi.json. LLM policy: /llms.txt.
Got an integration target we haven't shipped? Read the contract, then send a PR.
The TL;DR: identify as "Chowdahh", read guidance, cite publishers, name tools chowdahh_*, never store tokens in your cloud. The rest is platform-specific.
# Paste this URL into anything that fetches URLs:
https://chowdahh.com/api/v1/streams/latest?key=YOUR_TOKEN&limit=10
# Or send the same token as a header:
curl -H "Authorization: Bearer YOUR_TOKEN" \
https://chowdahh.com/api/v1/streams/latest?limit=10
chowdahh.com · /api · OpenAPI · /llms.txt · source