sec-verify-fix
Pass
Audited by Gen Agent Trust Hub on Nov 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script (
python scripts/secfind.py validate findings.json coverage.json). This script is expected to be present in the local environment as part of thesec-scanproject. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data including vulnerability findings (
findings.json) and security patches (fix.patchFile). An attacker could potentially embed malicious instructions or logic within these files to influence the verification outcome or bypass security checks. - Ingestion points:
findings.json,coverage.json, and patch files specified in the findings. - Boundary markers: No explicit instruction-ignoring delimiters or boundary markers are defined for the processed content.
- Capability inventory: File system read/write (patching), shell command execution (running tests), and local script execution (
scripts/secfind.py). - Sanitization: No sanitization or validation of the patch file content or finding evidence is mentioned before processing.
Audit Metadata