agent-self-recovery
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a standard utility for agent self-recovery and does not perform any risky or unauthorized operations.
- [COMMAND_EXECUTION]: The skill contains no shell commands, scripts, or instructions to execute external binaries.
- [EXTERNAL_DOWNLOADS]: There are no references to external URLs, remote scripts, or third-party package dependencies.
- [DATA_EXFILTRATION]: The skill does not access sensitive system files (such as credentials or SSH keys) and performs no network operations.
- [INDIRECT_PROMPT_INJECTION]: The skill manages its own state file to track recovery attempts. Because it processes the agent's own execution history rather than untrusted external data, it does not present an indirect injection surface.
Audit Metadata