lilbee

Beta — feedback and bug reports welcome. Open an issue.

Interactively or programmatically chat with a database of documents using strictly your own hardware, completely offline. Augment any AI agent via MCP or shell — take a free model or even a frontier model and make it better. Talks to an incredible amount of data formats (see supported formats). Integrate document search into your favorite GUI using the built-in REST API — no need for a separate web app when you already have a preferred GUI (see Obsidian plugin).

---

• Why lilbee
• Demos
• Install
• Quick start · Full usage guide
• Agent integration
• HTTP Server · API reference
• Interactive chat
• Supported formats

---

Why lilbee

• Your hardware, your data — chat with your documents completely offline. No cloud, no telemetry, no API keys required
• Make any model better — augment any AI agent via MCP or shell with hybrid RAG search. Take a free model or even a frontier model and make it leagues better at your data
• Talks to everything — PDFs, Office docs, spreadsheets, images (OCR), ebooks, and 150+ code languages via tree-sitter
• Bring your own GUI — built-in REST API means you can integrate document search into whatever tool you already use. No extra app needed (see Obsidian plugin)
• Per-project databases — lilbee init creates a .lilbee/ directory (like .git/) so each project gets its own isolated index

Add files (lilbee add), then search or ask questions. Once indexed, search works without Ollama — agents use their own LLM to reason over the retrieved chunks.

Demos

Click the ▶ arrows below to expand each demo.

AI agent — lilbee search vs web search (detailed analysis)

opencode + minimax-m2.5-free, single prompt, no follow-ups. The Godot 4.4 XML class reference (917 files) is indexed in lilbee. The baseline uses Exa AI code search instead.

⚠️ Caution: minimax-m2.5-free is a cloud model — retrieved chunks are sent to an external API. Use a local model if your documents are private.

	API hallucinations	Lines
With lilbee (code · config)	0	261
Without lilbee (code · config)	4 (~22% error rate)	213

With lilbee — all Godot API calls match the class reference

Without lilbee — 4 hallucinated APIs (details)

If you spot issues with these benchmarks, please open an issue.

Vision OCR

Scanned PDF → searchable knowledge base

A scanned 1998 Star Wars: X-Wing Collector's Edition manual indexed with vision OCR (LightOnOCR-2), then queried in lilbee's interactive chat (qwen3-coder:30b, fully local). Three questions about dev team credits, energy management, and starfighter speeds — all answered from the OCR'd content.

See benchmarks, test documents, and sample output for model comparisons.

One-shot question from OCR'd content

The scanned Star Wars: X-Wing Collector's Edition guide, queried with a single lilbee ask command — no interactive chat needed.

Standalone

Interactive local offline chat

[!NOTE] Entirely local on a 2021 M1 Pro with 32 GB RAM.

Model switching via tab completion, then a Q&A grounded in an indexed PDF.

Code index and search

Add a codebase and search with natural language. Tree-sitter provides AST-aware chunking.

JSON output

Structured JSON output for agents and scripts.

Hardware requirements

When used standalone, lilbee runs entirely on your machine — chat with your documents privately, no cloud required.

Resource	Minimum	Recommended
RAM	8 GB	16–32 GB
GPU / Accelerator	—	Apple Metal (M-series), NVIDIA GPU (6+ GB VRAM)
Disk	2 GB (models + data)	10+ GB if using multiple models
CPU	Any modern x86_64 / ARM64	—

Ollama handles inference and uses Metal on macOS or CUDA on Linux/Windows. Without a GPU, models fall back to CPU — usable for embedding but slow for chat.

Install

Prerequisites

• Python 3.11+
• Ollama — the embedding model (nomic-embed-text) is auto-pulled on first sync. If no chat model is installed, lilbee prompts you to pick and download one.
• Optional (for scanned PDF/image OCR): Tesseract (brew install tesseract / apt install tesseract-ocr) or an Ollama vision model (recommended for better quality — see vision OCR)

First-time download: If you're new to Ollama, expect the first run to take a while — models are large files that need to be downloaded once. For example, qwen3:8b is ~5 GB and the embedding model nomic-embed-text is ~274 MB. After the initial download, models are cached locally and load in seconds. You can check what you have installed with ollama list.

Install

pip install lilbee        # or: uv tool install lilbee

Development (run from source)

git clone https://github.com/tobocop2/lilbee && cd lilbee
uv sync
uv run lilbee

Quick start

See the usage guide.

Agent integration

lilbee can serve as a local retrieval backend for AI coding agents via MCP or JSON CLI. See docs/agent-integration.md for setup and usage.

HTTP Server

lilbee includes a REST API server so you can integrate document search into any GUI or tool:

lilbee serve                          # start on a random port (written to <data_dir>/server.port)
lilbee serve --port 8080              # or pick a fixed port

Endpoints include /api/search, /api/ask, /api/chat (with streaming SSE variants), /api/sync, /api/add, and /api/models. When the server is running, interactive API docs are available at /schema/redoc. See the API reference for the full OpenAPI schema.

Interactive chat

Running lilbee or lilbee chat enters an interactive REPL with conversation history, streaming responses, and slash commands:

Command	Description
/status	Show indexed documents and config
/add [path]	Add a file or directory (tab-completes paths)
/model [name]	Switch chat model — no args opens a curated picker; with a name, switches directly or prompts to download if not installed (tab-completes installed models)
/vision [name|off]	Switch vision OCR model — no args opens a curated picker; with a name, prompts to download if not installed; off disables (tab-completes catalog models)
/settings	Show all current configuration values
/set <key> <value>	Change a setting (e.g. /set temperature 0.7)
/version	Show lilbee version
/reset	Delete all documents and data (asks for confirmation)
/help	Show available commands
/quit	Exit chat

Slash commands and paths tab-complete. A spinner shows while waiting for the first token from the LLM. Background sync progress appears in the toolbar without interrupting the conversation.

Supported formats

Text extraction powered by Kreuzberg, code chunking by tree-sitter. Structured formats (XML, JSON, CSV) get embedding-friendly preprocessing. This list is not exhaustive — Kreuzberg supports additional formats beyond what's listed here.

Format	Extensions	Requires
PDF	.pdf	—
Scanned PDF	.pdf (no extractable text)	Tesseract (auto, plain text) or Ollama vision model (recommended — preserves tables, headings, and layout as markdown)
Office	.docx, .xlsx, .pptx	—
eBook	.epub	—
Images (OCR)	.png, .jpg, .jpeg, .tiff, .bmp, .webp	Tesseract
Data	.csv, .tsv	—
Structured	.xml, .json, .jsonl, .yaml, .yml	—
Text	.md, .txt, .html, .rst	—
Code	.py, .js, .ts, .go, .rs, .java and 150+ more via tree-sitter (AST-aware chunking)	—

See the usage guide for OCR setup and model benchmarks.

License

MIT