deeprepo 1.1.0

A production-grade Python library for RAG on local codebases with multi-provider AI support and MCP (Model Context Protocol) server integration

DeepRepo Core

A production-grade Python library for RAG on local codebases — code knowledge graph, hierarchical wiki engine, smart query routing, and branch isolation.

See the main README.md for complete documentation.

Quick Install

pip install deeprepo

# With MCP server support
pip install deeprepo[mcp]

Quick Start (CLI)

# Ingest current directory (Ollama, free and local)
deeprepo ingest .

# Browse the auto-generated wiki with in-page chat
deeprepo serve                     # → http://localhost:8080

# Ask questions directly
deeprepo query "how does authentication work?"
deeprepo query "what breaks if I change router.py?"

Quick Start (Python API)

from deeprepo import DeepRepoClient

# Initialize (defaults to Ollama if running, else OpenAI)
client = DeepRepoClient(provider_name="ollama")

# Ingest your code (incremental — unchanged files are skipped)
result = client.ingest("/path/to/your/code")
print(f"Indexed {result['files_scanned']} files, {result['wiki_generated']} wiki pages")

# Query with smart routing
response = client.query("How does authentication work?")
print(response['answer'])
print(f"Intent: {response['intent']}, Strategy: {response['strategy']}")
print(f"Sources: {response['sources']}")    # list of file paths

# Browse the wiki
print(f"Wiki at: {client.get_wiki_dir()}")

CLI Commands

Command	Description
deeprepo init	Detect provider setup, print the ingest command
deeprepo ingest [PATH]	Build graph + wiki + embeddings
deeprepo wiki [PATH]	Regenerate wiki pages only
deeprepo serve	Launch wiki viewer + chat at port 8080
deeprepo query "…"	Ask a question, get an AI answer
deeprepo status	Show branch isolation & cache freshness

What's Inside

Module	Purpose
client.py	Main facade — branch isolation, freshness, provider wiring
graph.py	SQLite store: graph nodes/edges, embeddings, wiki index, state
graph_builder.py	Tree-sitter AST parser → code knowledge graph
wiki.py	Hierarchical wiki engine — bottom-up LLM synthesis
router.py	Intent classifier + 6 context strategy selectors
ingestion.py	File scanner, chunker, language detection
ui.py	Wiki viewer (HTTP + mermaid renderer + chat)
mcp/server.py	7 focused MCP tools for AI assistants

MCP Tools (7 tools)

Connect to Cursor / Claude Desktop as an MCP server:

deeprepo-mcp     # starts the MCP server on stdio

Tool	Purpose
ingest_codebase	Index a repo directory
find_symbol	Locate a class/function (~50 tokens)
get_file_structure	File API without reading source (~150 tokens)
explain_file	Plain-English file explanation (~300 tokens)
find_change_impact	Blast-radius analysis (~300 tokens)
ask_codebase	Open-ended question about the code
get_project_overview	Whole-repo narrative overview

Design Diagrams

• High-Level Design — Process flow diagram
• Class Interaction Design — Class relationships

Open with excalidraw.com or the VS Code Excalidraw extension.

License

MIT License — see LICENSE file for details.