Documentation
RaggieCode Documentation
Everything you need to install, configure, and master RaggieCode .; from your first task to deep dives into the semantic code index, call graph traversal, subagent orchestration, and automatic context handover.
Getting Started
From zero to running RaggieCode on your project in under five minutes. Learn how to install via pip, run the interactive setup wizard to configure your LLM provider and API keys, navigate the interactive chat interface, and execute your first automated code change with full task planning and approval gates.
Architecture
Dive deep into how RaggieCode works under the hood. Understand the tree-sitter code indexing pipeline that parses 15+ languages, the SQLite symbol storage schema, the dependency graph with blast radius analysis via BFS traversal and cycle detection, the agent orchestration loop with tool dispatch, the subagent delegation system, and the built-in git undo/redo mechanism.
Agent Handover
Learn how RaggieCode preserves task state across LLM context window limits. Understand the handover trigger conditions, the structured handover document format (goal state, decisions, diffs, next step), the session chain for multi-handover tasks, and crash recovery. Includes a complete worked example from trigger to handoff.
Commands
Master RaggieCode's command-line interface. Covers all CLI commands (raggie setup, raggie code, raggie skill, raggie keys, raggie roles), in-chat commands (/undo, /redo, /effort, /help, /streaming, /reasoning, and more), and the five effort levels from Zen to Insane .; including how Zen still delegates through the todo list system in a linear, predictable way. Includes full argument reference and practical examples.
Configuration & Local AI
Configure RaggieCode for your workflow. Covers the ~/.config/raggie/ directory layout, roles.json fields for customizing agent behavior, AGENTS.md conventions, .aiignore enforcement, and full guides for running local models via Ollama, vLLM, and LM Studio .; including tool calling, context windows, and streaming.
Tools Reference
Complete catalogue of all 31 built-in tools grouped by category. Covers file operations, shell execution, code exploration (semantic index, call graph, symbol lookup), web search, todo lists, skills management, background services, and the subagent dispatch system. Includes notes on the tree-sitter code indexer that powers the code exploration tools.
Skills System
Extend RaggieCode's behaviour with persistent skills. Learn the full workflow: startup loading of role skills, on-demand fetching with GetSkill, persistence across sessions, multiple skills per role, user consent for SetSkill, override via AGENTS.md, and CLI management commands. Includes practical examples for creating custom behaviours.
Todo List System
Plan and track complex multi-step tasks. Covers the full workflow from GetActiveTodoList through CreateTodoList, AddTask, GetTodoList review, ApproveTodoList, and sequential ExecuteNextTask execution. Includes crash recovery patterns, nested todo lists for deep subtask delegation, and integration with the agent handover system.
Quick Links
Common pages and references to help you find what you need.
Getting Started
Install, setup, first task
Commands
CLI, in-chat, effort levels
Configuration
Roles, AGENTS.md, local AI
Architecture
Indexing, graph, internals
Agent Handover
Context preservation
Tools Reference
All 31 tools catalogued
Skills System
Persistent instructions
Todo List System
Plan, approve, execute
GitHub
Source code & issues
What is RaggieCode?
RaggieCode is an open-source AI coding agent built for large codebases. Unlike most AI coding tools that dump file contents into a prompt and hope for the best, RaggieCode builds a semantic code index using tree-sitter AST parsing across 15 languages, navigates call graphs with cycle detection and blast radius analysis, and performs automatic context handover when session limits are reached .; so complex, multi-step changes never lose progress.
7
Documentation guides
15
Supported languages
31
Built-in tools
# Install RaggieCode
$ pip install git+https://github.com/Hussein-L-AlMadhachi/RaggieCode.git
# Run the setup wizard
$ raggie setup
# Start coding (interactive mode)
$ raggie code /path/to/project
# Or single-shot mode
$ raggie code . --prompt "Add input validation to the login endpoint"Full walkthrough available in the Getting Started guide.