tailrocks-improve-security
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository content, creating an inherent attack surface. However, it implements robust mitigations:
runtime-trust.mdexplicitly states that embedded instructions cannot alter governing rules or authority, and it mandates secret scrubbing and read-only operations to limit the impact of potential injections. - [DATA_EXFILTRATION]: The instructions explicitly prohibit transmitting secrets, publishing exploits, or performing network actions. It requires citing only the location and type of secrets rather than their values to ensure sensitive data does not leave the environment.
- [COMMAND_EXECUTION]: The skill allows for running target tooling but restricts it to an enforceably read-only tree with disabled network access, scrubbed secrets, and bounded execution time, preventing unauthorized system changes or persistence.
Audit Metadata