# Your context, everywhere.

Give Claude, Cursor, ChatGPT, and other MCP clients direct access to your prompts, documents, and collections. One connection, 29 tools.

[Start free](https://contextrepo.com/pricing) · [Read the docs](https://contextrepo.com/docs/mcp)

29 tools · Streamable HTTP · stdio bridge v3

## Say it in plain language.

Your client picks the right tools behind the scenes. You never leave the conversation.

### "Use my PR review prompt on this diff."

Your client calls: `read_prompt`

Your client fetches the review template with all of its variables and applies it to the diff in front of you. Nothing to copy.

### "What did my research say about rate limiting?"

Your client calls: `deep_search`, `deep_read`

Deep Search finds the paragraphs that answer the question across every document, then reads the exact chunk so the answer comes with its source.

### "Save this conversation to my Architecture collection."

Your client calls: `create_document`, `add_to_collection`

Two tools, one sentence. The conversation becomes a versioned document and lands in the collection you named.

### "Roll back my API guidelines to the version before the rewrite."

Your client calls: `get_document_versions`, `restore_document_version`

Your client lists the recorded versions, picks the snapshot you meant, and restores it as a new revision. The versions in between stay in history.

## 29 tools. One URL.

Full CRUD, search, Deep Search, and a cited `reason` answer, over authenticated Streamable HTTP or the stdio bridge.

- **User** (1): `get_user_info`
- **Prompts** (7): `search_prompts`, `read_prompt`, `create_prompt`, `update_prompt`, `delete_prompt`, `get_prompt_versions`, `restore_prompt_version`
- **Documents** (7): `list_documents`, `get_document`, `create_document`, `update_document`, `delete_document`, `get_document_versions`, `restore_document_version`
- **Collections** (7): `list_collections`, `get_collection`, `create_collection`, `update_collection`, `delete_collection`, `add_to_collection`, `remove_from_collection`
- **Search** (1): `find_items`
- **ChatGPT App** (2): `search`, `fetch`
- **Deep Search** (3): `deep_search`, `deep_read`, `deep_expand`
- **Reasoning** (1): `reason`

## Search inside documents, not just titles.

Most tools stop at the filename. Deep Search lets your client search within document content, land on the exact paragraph, and walk the document like a table of contents without loading all of it.

- **Vector-powered content search.** Finds passages by meaning, not keyword. Ask a question and get the paragraphs that answer it, ranked.
- **Hierarchical navigation.** Move up to the parent section, down into paragraphs, or sideways through siblings. Five directions from any hit.
- **Session-aware deduplication.** Chunks already returned in the same session are filtered out, so a second search surfaces what the first one missed.

[Deep Search tools reference](https://contextrepo.com/docs/mcp/tools-reference)

## Two ways to connect.

Same account, same 29 tools either way.

### Point your client at one URL. (Remote / Streamable HTTP)

For clients that implement MCP OAuth or send custom authorization headers. No Node.js on your machine. Grok uses the dedicated /grok-mcp endpoint and sees the same tools.

1. Add https://contextrepo.com/mcp as a remote MCP server.
2. Sign in with your Context Repo account when the client asks, or set an Authorization: API-Key header.
3. Ask “List my prompts” to confirm the connection.

[Streamable HTTP setup](https://contextrepo.com/docs/mcp/setup-guides/other-clients)

### Run the open-source bridge. (Local / stdio bridge)

For clients that launch a local command. context-repo-mcp v3 forwards to the hosted endpoint, so tools and results stay identical to the remote path.

1. Create an API key under Dashboard › Settings › API Keys with the prompts.read and documents.read scopes.
2. Add npx -y context-repo-mcp with CONTEXTREPO_API_KEY to your client config (Node.js 18+).
3. Restart the client and ask “List my prompts”.

[Cursor and Claude Desktop guides](https://contextrepo.com/docs/mcp/setup-guides/cursor)

Working in a shell, a script, or CI? The CLI reaches the same prompts, documents, and collections through the REST API. [Get the CLI](https://contextrepo.com/cli)

## Your questions, answered.

### What is the Context Repo MCP Server?

The Context Repo MCP Server provides 29 tools for prompts, documents, collections, search, reasoning, and hierarchical retrieval. Clients must support authenticated Streamable HTTP or be able to launch the local stdio bridge.

### How do I install the Context Repo MCP Server?

Connect an OAuth-capable remote client to https://contextrepo.com/mcp. For a stdio-only host, add the context-repo-mcp v3 bridge with a Context Repo API key.

### What is Deep Search in Context Repo?

Deep Search lets AI assistants search within document content using vector similarity, not just titles. Results include hierarchical navigation (document > section > paragraph) so your AI can explore documents progressively without loading everything at once.

### Which authentication method should I use?

Remote clients that implement MCP OAuth sign in with your Context Repo account at https://contextrepo.com/mcp and carry your full authority. Clients that launch a local command use the stdio bridge with an API key from Dashboard > Settings > API Keys, limited to the scopes you grant the key. Both resolve to the same account.

### How does the server keep concurrent edits safe?

update_document and restore_document_version require an expectedRevision from get_document or the last acknowledgement. Each call makes one atomic write with no preflight read. A stale value returns REVISION_MISMATCH so your client can re-read and retry instead of overwriting someone else's change.

### Does it work with ChatGPT?

Yes. The hosted server exposes search and fetch aliases that follow the OpenAI Apps SDK Company Knowledge contract, so ChatGPT can cite your prompts, documents, and collections. Grok connects through the dedicated https://contextrepo.com/grok-mcp endpoint and sees the same 29 tools.

## Zero copy-paste. Full context. All 29 tools in every plan.

[Start free](https://contextrepo.com/pricing) · [context-repo-mcp on GitHub](https://github.com/Gitmaxd/context-repo-mcp)

## For agents

Remote clients need authenticated Streamable HTTP support; the npm v3 command bridges local stdio clients to the same hosted tool contract. All tools are included on every plan, and the 7-day free trial covers them too.

### 29 Tools Overview

#### User Info (1 tool)
| Tool | Description |
|------|------------|
| get_user_info | Get authenticated user info and API-key permission scopes |

#### Prompts (7 tools)
| Tool | Description |
|------|------------|
| search_prompts | Paginated prompt listing; optional search filters that fetched page by title or description |
| read_prompt | Get full prompt content including variables |
| create_prompt | Create a new prompt template |
| update_prompt | Update an existing prompt |
| delete_prompt | Permanently delete a prompt |
| get_prompt_versions | View version history of a prompt |
| restore_prompt_version | Restore a prompt to a previous version |

#### Documents (7 tools)
| Tool | Description |
|------|------------|
| list_documents | List a page of documents with titles |
| get_document | Get document content up to the REST response limit; use Deep Search for targeted retrieval from large documents |
| create_document | Create a new document |
| update_document | Required expectedRevision from get_document or the latest successful acknowledgement; exactly one REST write with no preflight GET |
| delete_document | Permanently delete a document |
| get_document_versions | View version history |
| restore_document_version | Required expectedRevision is the current document revision, not the target historical version; exactly one REST write with no preflight GET |

#### Collections (7 tools)
| Tool | Description |
|------|------------|
| list_collections | List a page of collections |
| get_collection | Get collection details and up to 50 member items |
| create_collection | Create a new collection |
| update_collection | Update collection metadata |
| delete_collection | Delete an empty collection; the collection must be empty, so remove all member items first |
| add_to_collection | Add prompts or documents to a collection |
| remove_from_collection | Remove items from a collection |

#### Search (1 tool)
| Tool | Description |
|------|------------|
| find_items | Semantic or literal search across prompts, documents, and collections (up to 10 matches per type). Literal mode searches titles, descriptions, and indexed document body text via hierarchical chunks — full-body literal matching, eventually consistent with chunking |

#### ChatGPT App (2 tools)
These tools provide ChatGPT App compatibility through the OpenAI Apps SDK Company Knowledge contract.

| Tool | Description |
|------|------------|
| search | ChatGPT App-compatible search returning citation-ready results |
| fetch | ChatGPT App-compatible fetch by id; full prompt content, document content up to the REST response limit, or collection metadata with up to 50 members, plus a citation URL. Use Deep Search for targeted retrieval from large documents |

#### Deep Search (3 tools)
| Tool | Description |
|------|------------|
| deep_search | Vector search within document content with hierarchical results |
| deep_read | Read a specific document chunk with full metadata |
| deep_expand | Navigate document hierarchy (up, down, next, previous, surrounding) |

#### Reasoning (1 tool)
| Tool | Description |
|------|------------|
| reason | Ask a question and get a best-effort cited answer with citations and non-exhaustive limitations. Disagreements appear as ordinary cited statements in the answer. (read-only) |

### Behavior notes

- **Version History**: Prompt content edits and changed canonical document snapshots create history; exact unchanged document saves are no-ops
- **MCP Document CAS**: update_document and restore_document_version each require expectedRevision. On REVISION_MISMATCH, read again; if the intended state is already present, stop, otherwise reapply against the latest content
- **Managed Editing**: Documents using lease_required must use the REST coordination workflow; MCP does not accept lease credentials
- **Semantic Search**: 1536-dimension vector search across all content
- **Session Deduplication**: Filters out chunks already returned in the same session; once matching chunks are exhausted, the search returns no results
- **Dual Authentication**: OAuth2 for interactive clients, API keys for automated access

A distinguishing Context Repo capability is hierarchical vector search inside documents: move up (parent), down (children), next/previous (siblings), or surrounding (context window) from any hit.

### Client configs

OAuth-capable clients should connect directly to the remote endpoint:

```
URL: https://contextrepo.com/mcp
Transport: streamable-http
Auth: OAuth2 (discovery: https://contextrepo.com/.well-known/oauth-authorization-server)
```

The `context-repo-mcp` v3 package is a network-dependent API-key stdio bridge to the same endpoint and passes through its tools and result shapes. Replace `gm_your_api_key_here` with a key from Dashboard > Settings > API Keys, then restart the client.

#### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "context-repo": {
      "command": "npx",
      "args": ["-y", "context-repo-mcp"],
      "env": {
        "CONTEXTREPO_API_KEY": "gm_your_api_key_here"
      }
    }
  }
}
```

Guide: [Cursor](https://contextrepo.com/docs/mcp/setup-guides/cursor)

#### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "context-repo": {
      "command": "npx",
      "args": ["-y", "context-repo-mcp"],
      "env": {
        "CONTEXTREPO_API_KEY": "gm_your_api_key_here"
      }
    }
  }
}
```

Guide: [Claude Desktop](https://contextrepo.com/docs/mcp/setup-guides/claude-desktop)

#### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "context-repo": {
      "command": "npx",
      "args": ["-y", "context-repo-mcp"],
      "env": {
        "CONTEXTREPO_API_KEY": "gm_your_api_key_here"
      }
    }
  }
}
```

Guide: [VS Code](https://contextrepo.com/docs/mcp/setup-guides/vs-code)

#### Streamable HTTP (`mcp-config.json`)

```json
{
  "mcpServers": {
    "context-repo": {
      "url": "https://contextrepo.com/mcp",
      "headers": {
        "Authorization": "API-Key gm_your_api_key_here"
      }
    }
  }
}
```

Guide: [Streamable HTTP](https://contextrepo.com/docs/mcp/setup-guides/other-clients)

### Grok

- Endpoint: https://contextrepo.com/grok-mcp
- Setup guide: https://contextrepo.com/docs/integrations/grok
- Authentication: hosted OAuth

### Resources

- npm: https://www.npmjs.com/package/context-repo-mcp
- GitHub: https://github.com/Gitmaxd/context-repo-mcp
- Documentation: https://contextrepo.com/docs
- API Key Setup: https://contextrepo.com/dashboard/settings
- Sign Up: https://contextrepo.com/pricing