Security Model

Security Model

Security practices, threat model, and controls.

---

Threat Model

Assets

• API Keys (Critical)
• Memory Content (High)
• Configuration (Medium)
• Embeddings (Medium)
• Task Data (Medium)

Threat Actors

• Malicious Plugins
• Prompt Injection
• Local Attacker
• Network Attacker

Security Controls

API Key Management

• Environment variable interpolation
• Never commit to version control
• Rotate keys periodically
• Use scoped keys

Database Security

• Embedded SQLite (no network)
• Parameterized queries
• File permissions (chmod 600)

Plugin Security

• Manual installation required
• Code review recommended
• No sandboxing (trust required)
• Plugin validation on load

MCP Security

• stdio transport (no network)
• Input validation (Zod schemas)
• Process-level isolation

Best Practices

# Secure config file
chmod 600 aistack.config.json

# Secure database
chmod 600 ./data/aistack.db

# Use environment variables
export ANTHROPIC_API_KEY="sk-..."

---

Related:

• SECURITY.md
• Configuration Guide