agent-autonomy-envelope
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns or security vulnerabilities were identified. The skill's primary purpose is to provide a framework for hardening AI agent execution environments and enforcing task-scoped restrictions.
- [COMMAND_EXECUTION]: The provided Python script
scripts/validate-envelope.pyis a standalone utility for validating JSON configurations. It uses standard library functions for JSON parsing and SHA-256 hashing. It does not execute arbitrary shell commands or process untrusted code. - [DATA_EXFILTRATION]: The autonomy envelope template and validation logic implement a 'deny-by-default' network policy and explicitly forbid access to sensitive host resources such as SSH agents, host environment variables, and the user's home directory.
- [PROMPT_INJECTION]: The skill instructions focus on defining structural and operational boundaries. They do not contain any patterns associated with instruction overrides, role-play jailbreaks, or safety filter bypasses.
- [REMOTE_CODE_EXECUTION]: The skill does not perform any network operations, downloads, or external package installations. All components are local and static.
Audit Metadata