@brainfork/brainfork-openclaw

Give your OpenClaw agents memory that lasts.

Every time your agent restarts, it forgets — your decisions, your architecture choices, your project context. You end up repeating yourself, correcting the same mistakes, re-explaining the same conventions.

Brainfork fixes that. This plugin connects OpenClaw to Brainfork, a sovereign memory layer that stores your agent's knowledge externally and recalls the right context before each session starts.

What You Get

🧠 Automatic context recall — relevant memories are injected before every agent run, so your agent picks up where it left off.

🔄 Workspace sync — your MEMORY.md and memory/**/*.md files are automatically pushed to Brainfork, keeping your knowledge base fresh without manual effort.

📋 Decision logging — every significant decision your agent makes is captured, tagged, and retrievable. No more "why did we choose Postgres over SQLite?" moments.

🔍 Hybrid search — keyword + vector similarity search across your entire knowledge base. Find anything, fast.

📦 Document management — push, archive, and organize documents through a clean API.

🔐 Sovereign by design — your data lives in a Brainfork server you control. Not locked into any model vendor. Portable, inspectable, yours.

Quick Start

# Install the plugin
openclaw plugins install @brainfork/brainfork-openclaw

# Run interactive setup
openclaw brainfork setup

The setup wizard offers two paths:

• Browser login (recommended) — opens your browser to authenticate with Brainfork. Best for desktops.
• Manual setup — prompts for your API URL, endpoint, and key. Best for headless servers and CI.

Then restart OpenClaw:

openclaw gateway restart

That's it. Your agents now have persistent memory.

How It Works

1. Before each session, the plugin queries Brainfork for context relevant to the current task and injects it into the agent's prompt.
2. During sessions, agents can search, fetch, and push documents — and log decisions with full context.
3. After sessions, workspace memory files are synced to Brainfork automatically.

Your agent stops starting from zero. It remembers what matters.

CLI Commands

openclaw brainfork setup     # Interactive setup (browser OAuth or manual)
openclaw brainfork index     # Sync workspace memory to Brainfork
openclaw brainfork status    # Check connectivity and sync state

Agent Tools

Once installed, your agents get access to these tools:

| Tool | What it does | |------|-------------| | brainfork_search | Keyword search across your knowledge base | | brainfork_vsearch | Semantic vector search for conceptual matches | | brainfork_query | Hybrid search (BM25 + vector) for best-quality results | | brainfork_fetch | Retrieve a full document by ID | | brainfork_push_document | Store new documents in Brainfork | | brainfork_get_decisions | Search logged decisions | | brainfork_log_decision | Record a decision with context and reasoning |

Configuration

The setup wizard handles configuration automatically. For manual configuration, add this to your openclaw.json under plugins.entries.brainfork-openclaw.config:

{
  baseUrl: "https://api.brainfork.ai",
  endpoint: "mcp/my-team",
  apiKey: "${BRAINFORK_API_KEY}",
  autoRecall: true,        // inject context before sessions
  autoIndex: true,         // sync workspace memory automatically
  captureDecisions: true,  // log agent decisions
  maxResults: 5,
  similarityThreshold: 0.2,
  maxTokens: 600,
  deleteMode: "archive",   // archive | delete | ignore
  requestTimeoutMs: 20000
}

Notes:

• endpoint — a full MCP URL or a path relative to baseUrl
• apiKey — a raw Brainfork key or a full Bearer ... / ApiKey ... header value
• deleteMode — what happens to remotely synced files when you delete them locally
• Sync state is stored under ~/.openclaw/memory/brainfork/

Local Development

From an OpenClaw source checkout:

cd extensions/brainfork-openclaw
npm install
npm run build
openclaw plugins install --link ./extensions/brainfork-openclaw

Rebuild after changes, then restart OpenClaw.

Troubleshooting

Browser OAuth doesn't open on a headless server — choose Manual setup instead, or pass credentials directly in config.

Firewall blocking localhost callback — the OAuth flow uses a temporary local server. Allow localhost on high ports, or use manual setup.

"Plugin installed but not configured" — run openclaw brainfork setup to configure authentication.

Token refresh failures — re-run openclaw brainfork setup to re-authenticate with fresh tokens.

Connection timeouts — increase requestTimeoutMs in config (default: 20,000ms, max: 120,000ms).

Pricing

Brainfork starts at €5/month with a 14-day free trial. Includes 25k requests and 1k indexed documents. Sign up at brainfork.is.

License

MIT