LLM Output
Overview
Every ciderpress site exposes three machine-readable interfaces out of the box:
These are generated at build time and served alongside your HTML pages. No configuration, no opt-in — they exist on every ciderpress site.
LLMs.txt
ciderpress follows the llms.txt specification to provide two endpoints that give LLMs and AI tools a structured view of your documentation.
/llms.txt — page index
A lightweight index of every page, grouped by section. Useful for discovery and navigation.
/llms-full.txt — full content
The complete text of every page, concatenated into a single document. Useful for feeding entire documentation sets into an LLM context window.
Per-page raw markdown
Every documentation page is also available as raw markdown by appending .md to the URL:
This is useful for:
- AI agents that prefer markdown over HTML
- Programmatic documentation consumers
- Downstream tooling that processes raw content
Example
Code
Output
OpenAPI pages
OpenAPI-generated pages replace the standard raw markdown button with a "Copy Markdown" button that produces a structured markdown representation of the operation, including method, path, parameters, request body, and responses.
Design Decisions
- Always on — machine-readable output requires no configuration. Every site gets it for free because discoverability by AI tools should be a default, not an afterthought.
- Build-time generation — LLMs.txt files are generated during
ciderpress build, not served dynamically. This keeps runtime simple and means the files are cacheable by CDNs. - Per-page
.mdendpoints — rather than forcing consumers to parse HTML or download the fullllms-full.txt, individual pages are available as raw markdown. This lets agents fetch only what they need.
Resources
References
- Content — how pages and sections define your site structure
- Configuration reference — full config field reference