🐦 Munin: Reliable Memory for Your OpenClaw Agent

Your agent forgets everything the moment you close the tab. You've seen the "pro" solutions—Mem0, Zep, LangMem—and you've seen their $20/month price tags.

We think that's insane.

Munin is the pragmatic alternative. It’s the "Reliable Friend" that sits on your agent’s shoulder and whispers: "Hey, remember when pa Kal said he preferred Python over Node last week?"

💰 Why Munin? (The "No-Brainer" Math)

That’s a 10x - 150x price gap. We aren't building "Enterprise Synergy Platforms". We're building a notebook for your AI.

🔒 Security & Privacy

Trust Statement:

Data stays where you tell it to stay. All data is encrypted locally (if E2EE is enabled) and isolated by your Context Core. We are completely blind to your plaintext.

External Endpoints:

Data is sent ONLY to your MUNIN_BASE_URL. We don't phone home. We don't sell your "memories" to train models.

💎 E2EE with GraphRAG (Elite Tier)

If your Munin project has E2EE with GraphRAG enabled, standard E2EE rules apply PLUS:

• You MUST generate a vector embedding of the plaintext content locally before encrypting.
• You MUST include the embedding array in the store payload: { "content": "encrypted...", "embedding": [0.1, -0.02, ...], "encryptionMeta": {...} }.
• Munin will use this vector for zero-knowledge semantic search. It will reject the request if the embedding is missing.

🚀 Setup (Fast & Honest)

1. Get your Munin key: Grab a free cloud key at munin.kalera.dev.
2. Configure OpenClaw: Add these to your .env or agent config:
• MUNIN_BASE_URL: Where your Munin lives (e.g., https://munin.kalera.dev).
• MUNIN_API_KEY: Your key to the kingdom.

(Note: Provide your agent with the Context Core ID in its system prompt so it knows which project to use when making tool calls).

1. Profit: Your agent now has long-term memory. No more repeating yourself.

Built with ❤️ by Kalera for the OpenClaw Ecosystem.