Agent Security Checklist — Free 5-Minute Safety Audit

Is your agent safe to run autonomously?

Most AI agents ship with zero security policy. No trust boundaries. No spending limits. No prompt injection defense. This free 13-point checklist lets you audit your agent's security posture in under 5 minutes.

What You Get

A structured checklist covering four critical areas:

1. Trust Boundaries — Are actions classified by risk? Is there a spending limit? Can the agent send external messages unsupervised?
2. Prompt Injection Defense — Does the agent treat external content as data? Is there a symmetry test? Are authority claims verified?
3. Session Security — Are API keys stored securely? Does the agent avoid outputting secrets? Is there an incident log?
4. Operational Safety — Boot sequence? Recovery procedure? Logging requirements?

Scoring

| Score | Rating | |-------|--------| | 12-13 | Production-ready | | 9-11 | Mostly safe — fix the gaps | | 5-8 | Risky — do not run unsupervised | | 0-4 | Dangerous — build security first |

Run the checklist. Fix the gaps. Ship with confidence.

Free — No strings attached. If you want the complete drop-in security framework with trust levels, attack vector defenses, and customization guides, check out the Agent Trust & Safety Framework ($9).