Baseline provides [llms.txt](https://llmstxt.org/) endpoints so that LLMs, AI agents, and other tools can consume our documentation programmatically.

## Endpoints

| Endpoint                                                                               | Description                                                                            |
| -------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------- |
| [`/llms.txt`](https://www.baseline.markets/llms.txt)                                   | Overview of Baseline, key concepts, FAQ, and a linked index of all documentation pages |
| [`/docs/llms-full.txt`](https://www.baseline.markets/docs/llms-full.txt)               | Full text of every documentation page in a single file                                 |
| [`/docs/{slug}.md`](https://www.baseline.markets/docs/contracts/cli.md)                | Markdown for a single doc page (preferred; returns 200)                                |
| [`/docs/{slug}.mdx`](https://www.baseline.markets/docs/contracts/cli.mdx)              | Same markdown body via legacy suffix                                                   |
| [`/docs/llms.mdx/{slug}`](https://www.baseline.markets/docs/llms.mdx/how-it-works/blv) | Internal processed markdown path (same content)                                        |
| [`/blog/llms.txt`](https://www.baseline.markets/blog/llms.txt)                         | Blog index with post titles, dates, and descriptions                                   |
| [`/blog/llms-full.txt`](https://www.baseline.markets/blog/llms-full.txt)               | Full text of every blog post                                                           |

## Usage

Start with `/llms.txt` for an overview and index. Use `/docs/llms-full.txt` to ingest all documentation at once, or fetch individual pages via `/docs/{slug}.md` when you only need specific topics.

You can also request the normal page URL with `Accept: text/markdown` to receive markdown without guessing a suffix.

All markdown endpoints return `text/markdown` with a 1-hour cache (`Cache-Control: public, max-age=3600`).