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

# For AI Agents

> Machine-readable access to Sentiment Data — the OpenAPI document, the Markdown index, and the Analyst Consensus MCP server (Streamable HTTP, no auth).

The Analyst Consensus API ships three machine-readable surfaces next to this documentation. They
are served by the consensus service itself, so they always describe the API as currently deployed.

## OpenAPI document

```text theme={null}
https://consensus.cryptoquant.com/openapi.json
https://consensus.cryptoquant.com/openapi.yaml
```

OpenAPI 3.1, version 2.2.0 — every endpoint with parameters, response schemas, validated examples,
the structured error body, every enum in full, and an `x-plan` block per operation saying who may
call it. Import it as a ChatGPT GPT Action (Authentication: None for the public operations) or any
OpenAPI tool loader. The same document is also served at `/.well-known/openapi.json`.

The spec also carries these top-level extension blocks:

| Block | Contents |
| - | - |
| `x-plans` | Every plan with its endpoints, rate and daily limits, history and raw-source-text rule |
| `x-raw-source-text` | How the Integration clearance works and how to request it |
| `x-errors` | Every error code with its HTTP status and recovery |
| `x-conventions` | Dates, pagination, filters, sorting, enums, caching |
| `x-mcp` | The MCP server URL, transport and tool list |
| `x-alternatives` | The CryptoQuant [Indicator API](/indicator-api/consensus) route to the daily index, `llms.txt`, `openapi.yaml` |

<Note>
  This documentation renders a snapshot of that document (`openapi/sentiment.json`). If the two ever
  disagree, the live `/openapi.json` is the source of truth.
</Note>

## Markdown index

```text theme={null}
https://consensus.cryptoquant.com/llms.txt
```

The Markdown index for answer engines — the same field reference as these pages. Append `.md` to
any page on this site as well (for example `/sentiment/consensus-index.md`).

## MCP server

```text theme={null}
https://consensus.cryptoquant.com/mcp
```

The same data as a remote **MCP server** — Streamable HTTP, read-only, stateless, **no
authentication**, per-IP rate limit. Point a Claude connector, ChatGPT or any MCP client at the URL
above. Tools:

| Tool | Backed by |
| - | - |
| `get_consensus` | [`GET /consensus`](/sentiment/consensus-index) — the Analyst Consensus Index for an asset |
| `get_analyst_views` | [`GET /analyst-views`](/sentiment/analyst-views) — what analysts are saying about an asset |
| `get_top_analysts` | [`GET /analysts/top`](/sentiment/top-analysts) — the most accurate analysts on an asset |
| `get_analyst_profile` | [`GET /analysts/{handle}`](/sentiment/analyst-track-record) — one analyst's track record |
| `get_recent_calls` | [`GET /calls`](/sentiment/calls) — the newest analyst calls on an asset |

Tag your surface with `?via=chatgpt` or `?via=claude` — the links in every answer carry the
matching `utm_source`. A plain `GET` on the URL (without an MCP `Accept: text/event-stream`)
returns a text description of the server.

<Info>
  This MCP server covers Sentiment Data only. The CryptoQuant on-chain and market data MCP server is
  at `https://mcp.cryptoquant.com/mcp` — see the [MCP Server guide](/guides/mcp-server).
</Info>

## Attribution

Every public payload carries `page_url` (link it), `cite_as` and `data_by`
("Data by CryptoQuant Consensus") — show the attribution with the numbers.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.