auto-legion
Pass
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection.
- Ingestion points: Reads project files in the stream scope (SKILL.md) to perform pre-analysis and craft agent prompts.
- Boundary markers: Uses markdown headers in templates but lacks explicit instructions to ignore embedded commands in source data.
- Capability inventory: Dispatches background sub-agents via the Agent tool and executes shell commands for verification.
- Sanitization: No sanitization is performed on data read from project files before prompt interpolation.
- [COMMAND_EXECUTION]: Executes shell commands for code verification between execution waves.
- Evidence: Usage of npx tsc --noEmit and npx vitest run to gate progress between phases.
Audit Metadata