🧶 Ragdoll

Overview

Ragdoll (Retrieval-Augmented Generation Driven by Offline Local LLMs) is a fully-local RAG system designed for engineering teams who need to search, summarize, and reason over internal knowledge sources — JIRA tickets, PDF documentation, and Python source code — without sending data to external services.

Note

Ragdoll is an independent open-source project created and maintained in a personal capacity. It does not represent the official software, technical roadmap, or endorsement of any employer or institution.

🧭 Where should I go?

• Just want to try it out? Head to the Getting Started guide.

• Ready to search, summarize, or chat? See Searching and Chatting and the Web UI.

• Want to connect Ragdoll to Claude/VS Code? Check out the MCP Integration.

• Need to ingest Jira, Bitbucket, GitHub, or Git? Read about Ingesting Data, Local Knowledge Staging, and Configuration.

• Want to build on top of Ragdoll? Dive into the System Architecture, API Reference, or learn about Extending Ragdoll.

Key Features

• Multi-source ingestion — PDF, JIRA, Bitbucket, GitHub, Git, and Python code

• Live Database Querying — Automatic Intent Routing between ChromaDB vector search and real-time Jira JQL, GitHub Search, and Bitbucket APIs

• Semantic search — ChromaDB vector store with cosine similarity and metadata filtering

• Local LLM — Ollama-powered embedding and generation

• Interactive chat — Multi-turn RAG chat with persistent history and prompt grounding

• Privacy-first — Everything runs locally; no external API calls

• Flexible configuration — 4-layer precedence (env → project → user → defaults)