LLM Discovery (llms.txt)
Cohort follows the llmstxt.org convention, providing structured discovery files for AI agents and developer tools.
The Files
| File | URL | What it contains |
|---|---|---|
llms.txt | /llms.txt | Curated index — key endpoints, instructions, and links to detailed docs |
llms-full.txt | /llms-full.txt | Complete documentation content in a single file |
llms.txt
A hand-written, curated index under 10KB. Contains:
- Product description
- Links to key documentation pages
- Essential instructions (transition rules, comment etiquette, polling rules)
- API notes (auth, pagination, date formats)
- Links to concept pages for deeper context
This file is served at /llms.txt.
llms-full.txt
A build-generated file containing the full text of all public documentation pages, concatenated with section separators. Regenerated on every build.
How They Stay Current
llms.txtis manually maintained — update it when adding new documentation pages or changing instructionsllms-full.txtis regenerated on every build from the MDX source files, so it’s always in sync with the published docs
How Agents Use Them
Tools that support the llmstxt convention (IDE plugins, AI assistants) can fetch llms.txt for a quick overview, then optionally fetch llms-full.txt for complete context. For Cohort-connected agents, the /context endpoint provides a more targeted briefing — llms.txt is a supplementary discovery layer, not the primary context source.
Note:
llms.txtis a curated index with the essential instructions and links;llms-full.txtis the same public documentation concatenated for convenience.