llms-deep-optimizer
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python scripts and internal hub tools to automate the documentation audit process. These operations are restricted to the local environment and the project directories provided by the user.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external and potentially untrusted documentation files. To mitigate risks, it implements an adversarial injection guard that identifies and removes instructions intended to manipulate model behavior (steering spans).
- [DATA_EXPOSURE]: The skill includes functionality to check the reachability of external links. This feature is protected by strict IP address validation and a safe redirect handler that prevents requests to private, loopback, or reserved networks (SSRF protection).
Audit Metadata