The docs for your agent
How an agent reads this documentation: the docs_search and docs_read tools in the Fransys MCP, llms.txt, every page in Markdown, and the search index.
Three ways to read the docs
Section titled “Three ways to read the docs”Your agent reads the same documentation as you, without the menu or the layout:
| Address | What it contains |
|---|---|
https://docs.fransys.cloud/en/llms.txt | The table of contents: one line per page, with its description and its Markdown link. |
https://docs.fransys.cloud/en/llms-full.txt | All the docs in a single file. |
https://docs.fransys.cloud/en/<page>.md | One page in Markdown. Add .md to a page’s address, without the trailing slash. |
The French version is at the root (https://docs.fransys.cloud/llms.txt, and so on).
On every page, Copy for an agent copies its Markdown version, and Open in Claude or Open in ChatGPT starts a conversation from the page.
Tell your agent
Section titled “Tell your agent”Add a line to your project’s instructions (CLAUDE.md, AGENTS.md, Cursor rules):
For anything about Fransys, first read https://docs.fransys.cloud/en/llms.txt,then the matching .md page. Cite the page you use.The search index
Section titled “The search index”https://docs.fransys.cloud/en/search-index.json splits each page into sections (one per level-2 heading), with their text in Markdown, their link and the date the page was checked.
In the Fransys MCP
Section titled “In the Fransys MCP”An agent connected to the Fransys MCP searches the docs with nothing to add:
docs_searchfinds the sections that answer a question;docs_readreads a page or a section in Markdown.
It relies on the search index above, and cites the page it read. To connect the MCP, see Connect an agent. An agent without the MCP goes through llms.txt.
The checked date
Section titled “The checked date”Each page shows when it was last checked against the console. A page marked Soon describes a feature that isn’t available yet.