@answer-engine/mcp-server

Model Context Protocol (MCP) server for Answer Engine — gives AI assistants like Claude access to your content library.

Install

npm install -g @answer-engine/mcp-server

Quick Start

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "answer-engine": {
      "command": "answer-engine-mcp",
      "env": {
        "ANSWER_ENGINE_API_KEY": "ae_live_your_key_here",
        "ANSWER_ENGINE_API_URL": "http://localhost:5050"
      }
    }
  }
}

Config file locations:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Cursor

Add to your Cursor MCP settings (.cursor/mcp.json):

{
  "mcpServers": {
    "answer-engine": {
      "command": "answer-engine-mcp",
      "env": {
        "ANSWER_ENGINE_API_KEY": "ae_live_your_key_here",
        "ANSWER_ENGINE_API_URL": "http://localhost:5050"
      }
    }
  }
}

npx (no install)

{
  "mcpServers": {
    "answer-engine": {
      "command": "npx",
      "args": ["-y", "@answer-engine/mcp-server"],
      "env": {
        "ANSWER_ENGINE_API_KEY": "ae_live_your_key_here",
        "ANSWER_ENGINE_API_URL": "http://localhost:5050"
      }
    }
  }
}

Configuration

| Environment Variable | Description | Default | |---|---|---| | ANSWER_ENGINE_API_KEY | Your Answer Engine API key (required) | — | | ANSWER_ENGINE_API_URL | API base URL | http://localhost:5050 |

Tools

The MCP server exposes 7 tools that AI assistants can call:

search_content

Search your content library using fulltext, semantic, or hybrid search.

| Parameter | Type | Required | Description | |---|---|---|---| | query | string | Yes | Search query text | | searchType | enum | No | hybrid (default), fulltext, semantic | | limit | number | No | Max results (1-50, default 10) | | include | string[] | No | Fields: summary, content, tags, metadata | | filters.contentTypes | string[] | No | Filter by content types | | filters.tags | string[] | No | Filter by tag slugs | | filters.dateFrom | string | No | ISO date — created after | | filters.dateTo | string | No | ISO date — created before |

Cost: 1 credit

get_content

Retrieve specific content items by ID.

| Parameter | Type | Required | Description | |---|---|---|---| | ids | string[] | Yes | Content item UUIDs (1-50) | | include | string[] | No | Fields: summary, content, tags, metadata, children, analysis |

Cost: 1 credit

list_tags

List all available tags, content types, and capabilities. Useful for discovering what's in your library before searching.

No parameters required.

Cost: Free (0 credits)

summarize_collection

Use AI to summarize or analyze content from your library.

| Parameter | Type | Required | Description | |---|---|---|---| | prompt | string | Yes | What to summarize or analyze | | filter.contentTypes | string[] | No | Filter by content types | | filter.tags | string[] | No | Filter by tag slugs | | filter.dateFrom | string | No | ISO date filter | | filter.dateTo | string | No | ISO date filter | | limit | number | No | Max items to analyze (1-100, default 20) |

Cost: 10 credits

scrape_url

Scrape a web page and save it to your content library.

| Parameter | Type | Required | Description | |---|---|---|---| | url | string | Yes | URL to scrape | | includeHtml | boolean | No | Include raw HTML (default false) |

Cost: 1 credit

crawl_domain

Crawl multiple pages from a domain.

| Parameter | Type | Required | Description | |---|---|---|---| | domain | string | Yes | Domain to crawl (e.g., example.com) | | maxPages | number | No | Max pages (1-100, default 10) |

Cost: 1 credit per page

web_search

Search the web and optionally scrape full content from results.

| Parameter | Type | Required | Description | |---|---|---|---| | query | string | Yes | Search query | | site | string | No | Limit to specific site | | scrapeResults | boolean | No | Scrape content from result URLs (default false) |

Cost: 1 credit

Resources

The server also exposes 2 MCP resources:

| Resource | URI | Description | |---|---|---| | Content Schema | answer-engine://schema | Content types, tags, capabilities, date range | | Content Item | answer-engine://content/{id} | Full content item by ID |

Error Handling

Errors are returned with isError: true in the MCP response with clear messages:

| HTTP Status | Error Message | |---|---| | 401 | Authentication failed. Check your ANSWER_ENGINE_API_KEY. | | 402 | Insufficient credits. | | 429 | Rate limit exceeded. |

Every successful response includes credit usage:

Credits charged: 1 | Credits remaining: 99

Development

# Clone the repo
git clone https://github.com/the-answerai/answer-engine.git
cd answer-engine

# Install and build
pnpm install
pnpm mcp-server:build

# Test with stdio
echo '{"jsonrpc":"2.0","method":"tools/list","id":1}' | \
  ANSWER_ENGINE_API_KEY=your_key node packages/mcp-server/dist/index.js

# Run tests
cd packages/mcp-server && pnpm test

License

MIT