# **not.bot MCP Server**

The not.bot MCP server gives AI assistants a public, read-only way to retrieve the published not.bot documentation with canonical citations. It is intended for answering questions about not.bot, not.bot Verify, not.bot Sign My Work, honest.bot, did:julia, credentials, claims, and the public product status described on this site.

## **Endpoint and transport**

Endpoint: `https://not.bot/mcp`

Transport: Model Context Protocol Streamable HTTP.

Server metadata: `https://not.bot/mcp/server.json`

## **Scope**

The MCP corpus contains the docs that are published in the website documentation manifest. Each response includes canonical HTML URLs and raw Markdown URLs so an assistant can cite the public page it used and, when useful, inspect the exact source Markdown.

Canonical citation behavior:

- Cite the canonical HTML URL for user-facing answers.
- Use the raw Markdown URL when a client needs the clean source document.
- Prefer section-level URLs when a tool result names a section.
- Do not cite the MCP transport URL as the source of a product claim unless the claim is about the MCP server itself.

## **Resources**

The server exposes these resource URI patterns:

- `notbot://docs/<slug>` for a complete published document.
- `notbot://docs/<slug>#<section-id>` for a section inside a published document.
- `notbot://glossary/<term>` for an individual glossary entry.

## **Tools**

The server exposes these tools:

- `search_docs(query, product?, audience?, limit?)`: Search the public docs and return ranked results with canonical citations.
- `get_doc(slug)`: Retrieve one published document by slug.
- `get_glossary_term(term)`: Retrieve one glossary entry by term.
- `get_product_status()`: Retrieve a compact public product-status summary.
- `get_verify_claims_catalog()`: Retrieve the public not.bot Verify claims catalog.

## **Example client configuration**

```json
{
  "mcpServers": {
    "notbot": {
      "url": "https://not.bot/mcp"
    }
  }
}
```