bagman

# Bagman Secure key management patterns for AI agents handling private keys and secrets. Designed to prevent: - **Key loss**: Agents forgetting keys between sessions - **Accidental exposure**: Keys leaked to GitHub, logs, or outputs - **Prompt injection**: Malicious prompts extracting secrets ## Core Principles 1. **Never store raw private keys in config, env vars, or memory files** 2. **Use session keys / delegated access instead of full control** 3. **All secret access goes through 1Password CLI (`op`)** 4. **Validate all outputs before sending to prevent key leakage** ## References - `references/secure-storage.md` - 1Password patterns for agent secrets - `references/session-keys.md` - ERC-4337 delegated access patterns - `references/leak-prevention.md` - Pre-commit hooks and output sanitization - `references/prompt-injection-defense.md` - Input validation and output filtering --- ## Quick Reference ### DO ✅ ```bash # Retrieve key at runtime via 1Password PRIVATE_KEY=$(op read "op://Agents/my-agent-wallet/private-key") # Use environment injection (key never touches disk) op run --env-file=.env.tpl -- node agent.js # Use session keys with bounded permissions # (delegate specific capabilities, not full wallet access) ``` ### DON'T ❌ ```bash # NEVER store keys in files echo "PRIVATE_KEY=0x123..." > .env # NEVER log or print keys console.log("Key:", privateKey) # NEVER store keys in memory/journal files # Even in "private" agent memory - these can be exfiltrated ...