> ## Documentation Index
> Fetch the complete documentation index at: https://docs.reply.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Machine-readable indexes

> llms.txt, OpenAPI, catalog.json, and runtime discovery — how agents read Reply.io documentation programmatically.

Every layer of Reply.io documentation is available in a machine-readable form. Use these instead of
scraping HTML.

## This site (docs.reply.io)

| Resource      | URL                                                     | What it is                                                                 |
| ------------- | ------------------------------------------------------- | -------------------------------------------------------------------------- |
| llms.txt      | [`/llms.txt`](https://docs.reply.io/llms.txt)           | Index of all documentation pages with descriptions                         |
| llms-full.txt | [`/llms-full.txt`](https://docs.reply.io/llms-full.txt) | The full documentation content in one file                                 |
| OpenAPI spec  | see [OpenAPI & Postman](/openapi-postman)               | The complete v3 REST API specification (also drives the API Reference tab) |

Mintlify also serves any docs page as markdown — append `.md` to a page URL for the raw source.

## The agents portal (agents.reply.io)

The [agent discovery portal](https://agents.reply.io) publishes its own machine surface, useful as
a compact capability index before diving into full references here:

| Resource                       | What it is                                                                               |
| ------------------------------ | ---------------------------------------------------------------------------------------- |
| `/catalog.json`                | Machine-readable module index: modules, workflows, auth model, rate limits, MCP endpoint |
| `/llms.txt` · `/llms-full.txt` | Portal content index and full text                                                       |
| `.md` twins                    | Every portal page is also plain markdown — append `.md` to its URL                       |
| Per-module OpenAPI             | Small, module-scoped OpenAPI subsets — lighter than the full spec for narrow tasks       |

The portal is a discovery layer; canonical technical facts live here on docs.reply.io.

## Runtime discovery

Two sources are always current because they come from the running systems:

* **MCP `tools/list`** — the live tool catalog with full JSON schemas and safety annotations.
  This is the runtime truth for MCP; regenerate the reference any time — see
  [Tool reference](/mcp/tools#regenerate-this-reference).
* **CLI `--help`** — `reply --help` and `reply <command> --help` print flags and examples, matching
  the installed version exactly.

## Recommended agent reading order

1. `docs.reply.io/llms.txt` — find the relevant pages.
2. The specific page as `.md` — read only what the task needs.
3. For MCP work: `tools/list` on the live server for exact schemas.
4. For REST work: the [OpenAPI spec](/openapi-postman) for exact request/response shapes.