Every page of this documentation has a raw markdown twin, and the whole set has a machine-readable index. Read the index to see what exists, then fetch pages as markdown.

## The endpoints

| URL | Returns |
|---|---|
| `https://www.monofile.ai/docs/llms.txt` | The index of every published page |
| `https://www.monofile.ai/docs.md` | The same index, under a second name |
| `https://www.monofile.ai/docs/<slug>.md` | One page's markdown, exactly as stored |
| `https://www.monofile.ai/docs/llms-full.txt` | Every published page concatenated into one file, for readers that want the whole corpus in a single fetch |

All of them respond with `Content-Type: text/markdown; charset=utf-8`. `https://www.monofile.ai/docs/index.md` is another alias for the index.

## Start with the index

```bash title="Terminal"
curl -s https://www.monofile.ai/docs/llms.txt
```

The shape is an H1, a quoted one-line summary, a flat list of the pages that sit at the root, then one `## Heading` per top-level section with its pages under it:

```md nocopy
# Monofile Documentation

> Monofile is a markdown workspace and cloud filesystem that people and AI agents share. Every page below has a raw markdown twin at the linked URL; /docs/llms-full.txt is every page concatenated.

- [Introduction](https://www.monofile.ai/docs/introduction.md): What Monofile is: a markdown workspace and cloud filesystem that people and agents share.

## Getting Started

- [Quickstart](https://www.monofile.ai/docs/getting-started/quickstart.md): Create your first file, edit it, and share it in five minutes
- [Agent Quickstart](https://www.monofile.ai/docs/getting-started/agent-quickstart.md): Register an agent identity, get an API key, and make your first tool call
```

What the index guarantees:

- **It is generated from the live documentation.** If a page is not in the index, it does not exist or is not published yet.
- **Order is the sidebar order.** Pages appear by section order, then by their order inside the section, so top to bottom is the intended reading sequence.
- **Every link is absolute and correct for the host you asked.** Request the index from a different environment and the links point at that environment.
- **Only published pages appear.** Section folders are headings, not entries, because a folder has no content of its own.
- **Each bullet carries the page's one-line description**, which is usually enough to decide whether to fetch it.

## Fetch a page

Take any `.md` URL from the index and get it verbatim:

```bash title="Terminal"
curl -s https://www.monofile.ai/docs/getting-started/core-concepts.md
```

What comes back is the stored markdown, with no HTML, no wrapper and no frontmatter. The body does not contain the page title. Read the title from the index.

The markdown uses GitHub-flavoured conventions plus two things to handle:

- `[[Page Title]]` and `[[Page Title|display text]]` are cross-references between pages, resolved by title. To follow one, match the title against the index and use that entry's `.md` URL.
- ` ```mermaid ` fenced blocks contain diagram source, and the source is usually more informative than the rendered picture.

> [!NOTE]
> A slug that does not exist, or an unpublished page, returns `404` with a JSON body. Published pages are cached for five minutes, so a change can take that long to appear.

## A note on scope

These endpoints serve the **documentation**. The API reference an agent works from day to day is the skill set at `https://www.monofile.ai/skill.md`, `api.md`, `mcp.md` and `skill.json`, described in [[Skill Files]]. Read the docs to understand the product and the skill files to call it. `GET /api/tools` is the authoritative list of what your key can do.

## Next steps

:::cards
- [[Limits]] — Sizes, counts, timeouts and rate limits in one table.
- [[Troubleshooting]] — The errors people actually hit, and the fix.
- [[Skill Files]] — The machine-readable onboarding set agents can fetch and follow.
- [[Agent Quickstart]] — Register an agent identity, get an API key, and make your first tool call.
:::