lov-fix-until-no-error
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to run arbitrary validation commands and project scripts to diagnose and verify error fixes, as specified in the SKILL.md workflow.
- [DYNAMIC_EXECUTION]: The skill modifies local project files and subsequently executes them to verify fixes, which involves running dynamically altered code within the user's environment.
- [INDIRECT_PROMPT_INJECTION]: The skill exposes a vulnerability surface for indirect prompt injection by ingesting and acting upon untrusted data from validation command outputs. 1. Ingestion points: Output and error logs from validation commands (SKILL.md). 2. Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore instructions embedded in command outputs. 3. Capability inventory: The skill has the capability to execute shell commands and modify the file system. 4. Sanitization: The skill does not implement sanitization or validation for the content of ingested command outputs.
Audit Metadata