Security and Safety

agentic security audit checklist

Comprehensive security audit checklist for agentic AI systems. Covers permission models, prompt injection, data protection, and compliance.

11 items
  1. 1
    implement permission models for all tools

    Permission Models

  2. 2
    enforce least privilege access

    Least Privilege

  3. 3
    test for prompt injection vulnerabilities

    Prompt Injection

  4. 4
    prevent data exfiltration paths

    Data Exfiltration

  5. 5
    review owasp top 10 for llms

    OWASP Top 10 for LLMs

  6. 6
    configure audit logging

    Audit Logging

  7. 7
    set rate limits on all endpoints

    Rate Limiting

  8. 8
    verify compliance requirements

    Compliance

  9. 9
    sandbox tool execution

    Tool Sandboxing

  10. 10
    contain blast radius of failures

    Blast Radius Containment

  11. 11
    use ephemeral execution environments

    Ephemeral Execution Environments

Related tracks

Tool Design and Contracts
Security and Safety