Gigabrain

Gigabrain is the local-first memory governance & trust control plane for people who run more than one AI agent. When you use Codex, Claude Code, Cursor/Windsurf, OpenClaw, and Hermes against the same work, each one quietly remembers different — and sometimes contradictory — things. Gigabrain gives you one inspectable place to see what they all remember, where each memory came from, what's stale, contradictory, or secret-leaking, and what is safe to hand to the next agent.

The product face is the Memory Passport: an audit report + safe handoff briefs. Underneath it is an event-sourced memory engine — SQLite-backed hybrid recall, deterministic dedupe/audit flows, read-only host-memory adapters, native markdown sync, a claim-slot world model with contradiction detection, and an optional web console.

Gigabrain does not replace native memories and does not claim to be a portable "carry your brain anywhere" bundle. It sits above native memory: it connects, checks, deduplicates, attributes, and audits the memory your agents already keep where the host exposes local files or you provide an explicit export, then writes safe briefs for the next host. Closed cloud memories (ChatGPT, Claude.ai, Gemini, Microsoft Copilot) are manual import/export only — no scraping, no hidden sync. Encrypted, signed, fully round-trippable export bundles are on the roadmap, not shipped today; see the destination audit.

Supported clients

Quickstart

I use multiple agents

npm install @legendaryvibecoder/gigabrain
npx gigabrain-codex-setup --project-root /path/to/repo
npx gigabrain-claude-setup --project-root /path/to/repo
npx gigabrain-hermes-setup --config ~/.gigabrain/config.json --workspace-root /path/to/repo
npx gigabrainctl sync-hosts --config ~/.gigabrain/config.json --host codex,claude_code,hermes,cursor,windsurf
npx gigabrainctl passport --config ~/.gigabrain/config.json --output-dir ./gigabrain-passport

This gives Codex, Claude, and Hermes the same local project/user memory store, indexes visible local host memories read-only, then writes a static Memory Passport report plus handoff briefs under ./gigabrain-passport/. See the Memory Passport guide, Codex setup, Claude setup, Hermes setup, the Nimbus Memory Bridge Contract, the destination audit, and the cross-memory pivot.

Manual cloud imports must be explicit:

npx gigabrainctl sync-hosts --config ~/.gigabrain/config.json \
  --manual-import ./chatgpt-memory-export.md \
  --manual-source-host chatgpt_manual
npx gigabrainctl passport --config ~/.gigabrain/config.json --output-dir ./gigabrain-passport

Codex only

npm install @legendaryvibecoder/gigabrain
npx gigabrain-codex-setup --project-root /path/to/repo
.codex/actions/verify-gigabrain.sh

Full setup guide

Claude / OpenClaw

npm install @legendaryvibecoder/gigabrain
npx gigabrain-claude-setup --project-root /path/to/repo
.claude/actions/verify-gigabrain.sh

For OpenClaw:

openclaw plugins install @legendaryvibecoder/gigabrain
cd ~/.openclaw/extensions/gigabrain && npm run setup -- --workspace /path/to/workspace
npx gigabrainctl doctor --config ~/.openclaw/openclaw.json

Full guides: Claude, OpenClaw

Hermes / Nimbus migration

npm install @legendaryvibecoder/gigabrain
npx gigabrain-hermes-setup --config ~/.gigabrain/config.json --workspace-root /path/to/workspace --install --test
hermes gateway restart

Import a legacy Nimbus/OpenClaw backup with provenance:

npx gigabrainctl import-openclaw \
  --config ~/.gigabrain/config.json \
  --registry /path/to/backup/clawd/memory/registry.sqlite \
  --source-label nimbus-backup \
  --dry-run

Full guide: Hermes setup Nimbus runtime policy: Nimbus Memory Bridge Contract

Upgrading? See the upgrade guide.

How it works

Conversation (OpenClaw / Codex / Claude Code / Claude Desktop)
               │
               ▼
┌──────────────────────────────────┐
│           Gigabrain              │
│   (memory layer + MCP server)   │
├──────────────────────────────────┤
│  Capture ─► Policy ─► Registry  │
│  Recall  ◄─ Orchestrator        │
│  Host Sync ◄─ Codex/Claude/etc. │
│  Native Sync ◄─► MEMORY.md      │
│  World Model (entities/beliefs) │
└──────────────┬───────────────────┘
               │
         SQLite + FTS5

Highlights

• Capture — hybrid model: explicit remember intent writes durable memory, Codex checkpoints write episodic session logs
• Recall — orchestrator chooses between quick context, entity briefs, timeline briefs, and verification-oriented recall automatically
• MCP tools — gigabrain_recall, gigabrain_remember, gigabrain_checkpoint, gigabrain_provenance, gigabrain_recent, gigabrain_sources, gigabrain_sync_status, gigabrain_export_brief, gigabrain_doctor, gigabrain_entity, gigabrain_contradictions, gigabrain_relationships
• Memory Passport — static Markdown/HTML source inventory, readiness verdict, dedupe audit, contradiction audit, stale-memory audit, provenance gaps, secret-risk flags, and safe handoff briefs that omit secret-risk rows entirely
• Host memory bus — read-only local sync for Codex, Claude Code, Cursor/Windsurf, OpenClaw native memory, and bridge-friendly Hermes integrations
• Dedupe — exact + hybrid semantic deduplication with type-aware thresholds
• Native sync — indexes MEMORY.md and daily notes alongside the registry for unified recall
• World model — entities, beliefs, episodes, open loops, contradictions, and syntheses
• Quality gate — junk filters, plausibility heuristics, and LLM second opinion
• Audit — nightly maintenance with snapshots, archive reports, and quality scoring
• Eval — in-process recall eval harness, nightly quality history, and latency summaries
• Web console — optional FastAPI dashboard for browsing and managing memories
• Person service — entity mention tracking for person-aware retrieval ordering

Key subsystems