Repowise is a free MCP server tool from Repowise Dev, built with Python 3.11+. It delivers codebase intelligence to developers and AI coding agents. This program indexes repository structure, Git history, tests, architectural decisions, documentation, and code-health signals. Repowise serves this pre-computed information to AI assistants, aiming to enhance their understanding and efficiency. It is distributed via pip, making it accessible for integration into development workflows.

What does Repowise connect to?

Repowise connects directly to a codebase, indexing its structure, Git history, tests, architectural decisions, documentation, and code-health signals. This process gathers essential intelligence about the repository. As an MCP server tool, Repowise then serves this collected information to AI coding agents. Its primary function is to provide a detailed, organized view of a software project, making complex codebases more understandable for both human developers and artificial intelligence.

Core features and tools

Repowise exposes callable tools through its CLI, including init, update, reindex, serve, watch, mcp, search, ask, context, symbol, why, status, distill, health, risk, workspace, capture, list, govern, add, remove, refresh, and uninstall. It also offers ten task-shaped MCP tools for AI assistants, such as get overview, get health, get risk, get_answer, and get_context.

Who is it for?

Repowise is for developers, AI coding teams using MCP-compatible agents like Claude Code or Cursor, and solo developers auditing complex or inherited codebases. Platform engineers tracking ownership and bus factor also benefit. It supports various AI clients including Windsurf, Codex, GitHub Copilot, and Cline, along with optional LLM providers like Anthropic, OpenAI, and Gemini models. Skip this if you do not work with codebases or AI coding agents.

How to get started

No credentials are required for the local MCP server, as it binds to 127.0.0.1 and uses the local socket for authentication. For features requiring an LLM, an API key such as ANTHROPIC_API_KEY, OPENAI_API_KEY, or GEMINI_API_KEY is needed. The MCP server uses streamable HTTP and Server-Sent Events (SSE) for transport.

Use cases

• Enhancing AI agent understanding and efficiency: AI agents use Repowise's pre-computed codebase intelligence via MCP tools, reducing file reads and lowering token usage.
• Codebase analysis and health assessment: Developers gain insights into complex software systems, identify critical components, detect dead code, and obtain code health scores.
• Automated documentation and architectural decision tracking: Repowise generates a structured wiki and captures architectural decisions, ensuring documentation remains accurate.

Repowise provides a specialized tool for developers and AI agents to understand and interact with codebases. It offers deep analysis, documentation generation, and support for various AI clients and LLM providers. Its local authentication model simplifies initial setup for analysis-only tasks. Start by running "repowise init" to index a repository and explore its structure.