draft-piece
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads and executes commands from the brief.md file, creating a surface for malicious instructions to be executed.
- Ingestion points: brief.md (Execution-checked code section).
- Boundary markers: No sanitization is performed; the brief is treated as a trusted contract.
- Capability inventory: Execution of shell commands.
- Sanitization: None.
- [COMMAND_EXECUTION]: The agent is explicitly instructed to run commands found in the repository's markdown files, which allows for arbitrary execution of local scripts and commands.
Audit Metadata