Your brain, persisted.
Zero cloud, zero keys.

A high-performance local knowledge engine that turns your files into a searchable, intelligent second brain — without ever leaving your machine.

Get Started View Source

Install

curl -fsSL https://raw.githubusercontent.com/macro88/gigabrain/main/scripts/install.sh | sh

Then build your first brain and connect it to Claude Code:

gbrain init ~/brain.db
gbrain import ~/notes --db ~/brain.db
GBRAIN_DB=~/brain.db gbrain serve

Three commands. No API keys. No cloud. Full install options →

Built for two audiences

For developers

Own your knowledge stack

One binary. One file. SQLite-backed full-text search and a local vector index, with a CLI you can drive from any shell. Ten minutes from install to a brain that answers your first question.

Build your first brain

For agents

MCP-native by design

Seventeen brain_* tools over stdio JSON-RPC. Same surface humans use; same data, same contracts. Privacy-safe gap logging, OCC-protected writes, and a sensitivity contract that keeps raw queries from ever leaving the machine by accident.

Open the agent quickstart

What makes GigaBrain different

Local-first

Compute and storage on your machine. The embedding model ships in the binary on the airgapped channel; on the online channel it caches once after a one-time download. Nothing leaves the box.

Agent-native

The MCP server predates any GUI. Same surface, same contracts, whether the caller is a person at the terminal or a model over JSON-RPC.

Single binary

No container, no Python runtime, no service mesh. Drop one statically-linked file on a laptop, a Pi, or an air-gapped workstation and it works the same.

Hybrid retrieval

FTS5 for precision and a local BGE vector index for semantic recall, fused with set-union and a short-match short-circuit. You don’t pick keyword or semantic — both run.

What you get out of the box

CLI commands

Read, write, graph, intelligence, collections, system.

MCP tools

Same surface, exposed to any MCP-compatible client.

Embedded skills

Ingest, query, maintain, enrich, briefing, alerts, research, upgrade.

Static binary

No daemon, no port, no Docker, no Python.

API keys required

The model runs on your CPU. Nothing transits your network.

Architecture for privacy

SQLite Everything

No complex setup. Your entire knowledge graph is stored in a highly optimized SQLite database on your local disk. Lightweight, portable, permanent.

Local Embeddings

Run state-of-the-art vector embeddings locally. Your private data never touches an API. Privacy-first AI that works completely offline.

Hybrid Search

Combine the precision of keyword search with the context of semantic search. Find exactly what you need across millions of documents in milliseconds.

Live Vault Sync

Attach any Obsidian vault as a collection. GigaBrain watches for edits with a 1.5 s debounce and reconcile-backed flushes so the brain stays current as you write. (Unix/macOS/Linux only — v0.9.6)

Get running in seconds

Three commands. No API keys. No cloud.

# Initialize a brain
gbrain init ~/brain.db

# Import your notes (idempotent — safe to re-run)
gbrain import ~/notes --db ~/brain.db

# Expose 17 tools to any MCP client over stdio (Unix/macOS/Linux; v0.9.6)
GBRAIN_DB=~/brain.db gbrain serve

Next: Build your first brain · Connect Claude Code · Why GigaBrain

Your brain, persisted. Zero cloud, zero keys.