lov-fix-general
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to run arbitrary shell commands to diagnose and verify software fixes, which is a powerful capability that allows it to interact deeply with the local system and project environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection due to its core workflow.\n
- Ingestion points: The skill reads untrusted data from external sources, including project source code, error logs, and environment configurations (SKILL.md Workflow).\n
- Boundary markers: The skill defines execution boundaries in SKILL.md, distinguishing between read-only analysis and destructive operations like modification, commit, and push.\n
- Capability inventory: It has the power to modify files, execute arbitrary shell commands, commit code to version control, and interact with a local profile persistence script (SKILL.md, scripts/profile_store.py).\n
- Sanitization: No specific evidence was found in the instructions regarding the sanitization or escaping of the ingested code or error content before it is processed by the model context.
Audit Metadata