Start with your codebase

Install the package, then run buddhi init from a repository. The command scans supported source files, writes a graph under .buddhi/graphs/, plans code documentation, and scaffolds an agent harness.

pip install "buddhi-ai[mcp]"
buddhi init

See the installation guide for pipx, uv, and platform notes.

What the CLI provides

• Code graph: tree-sitter extracts source structure and relationships. buddhi graph can report statistics, communities, cycles, paths, and change impact.
• MCP server: buddhi-mcp exposes code search, compressed file reading, memory, contrast checking, and accessibility verification tools over stdio.
• Code documentation: buddhi docs generates OKF documentation and can build or serve the generated Pelican site.
• Agent harness: buddhi init adds Antigravity workflows, specialist agents, rules, hooks, and templates to a project.
• Project memory: buddhi memory manages structured, source-aware concepts and SDD decisions.
• Developer utilities: buddhi scripts includes accessibility checks, anti-slop settings, output compression, and other repository tools.

Choose a guide

• CLI overview explains the graph, MCP server, and harness.
• Installation covers Python requirements and package managers.
• Commands lists the CLI command groups.
• Graph analysis explains graph inspection commands.
• Documentation generation covers OKF docs and the Pelican site.
• MCP server and tools documents the server interfaces.
• Project memory covers memory commands and concepts.
• Scripts and utilities describes the bundled utilities.
• Agent harness explains workflows, skills, rules, and hooks.