Inkeep

Inkeep

RAG Search over your content powered by Inkeep

22
Stars
10
Forks
0
Releases

Overview

Inkeep MCP Server enables retrieval-augmented generation (RAG) over your own documentation and product content using Inkeep. It is designed to be integrated with MCP clients such as Claude Desktop to provide question-answer capabilities that fetch relevant content from your workspace. The server runs locally and is managed with the uv Python project manager; local setup involves creating a venv and installing dependencies from pyproject.toml. To use, you must have an Inkeep account and generate an API key from the Inkeep Dashboard. The server is configured via environment variables: INKEEP_API_BASE_URL (default https://api.inkeep.com/v1), INKEEP_API_KEY, INKEEP_API_MODEL (e.g., 'inkeep-rag'), INKEEP_MCP_TOOL_NAME (e.g., 'search-product-content'), and INKEEP_MCP_TOOL_DESCRIPTION describing how to use the tool. An example claude_desktop_config.json shows how to add this MCP server under mcpServers with command 'uv', the project directory, and run arguments to start the inkeep_mcp_server module. The server queries the Inkeep API to retrieve product documentation and other content to answer conversational questions. This setup targets developers and teams seeking robust content search and RAG-based Q&A using Inkeep content.

Details

Owner
inkeep
Language
Python
License
MIT License
Updated
2025-12-07

Features

RAG-based content retrieval

Performs retrieval-augmented generation against your docs and product content using Inkeep's RAG model.

API-key authenticated access

Authenticates requests to the Inkeep API using a user-provided API key and base URL.

Model configurability

Configures the Inkeep model via INKEEP_API_MODEL (e.g., 'inkeep-rag'), enabling model switching.

Product content search tool

Exposes a dedicated MCP tool named 'search-product-content' to retrieve product documentation.

Environment-driven configuration

Uses environment variables to configure API base URL, key, model, tool name, and tool description for flexible deployment.

Local development setup with uv

Guides local setup with uv venv and dependency installation from pyproject.toml.

Claude Desktop integration

Designed to integrate with Claude Desktop MCP workflows via claude_desktop_config.json.

Run via uv command

Run the server with uv using the directory and module, e.g., uv run -m inkeep_mcp_server.

Audience

DevelopersSet up and run the MCP server to enable content search.
Product teamsLeverage Inkeep content to power product documentation search via MCP.
Claude Desktop usersIntegrate the MCP server into Claude Desktop workflows for RAG queries.

Tags

MCPInkeepRAGsearchdocumentationproduct contentAPIintegrationClaude Desktop