wizard
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads project files like READMEs and configuration files to determine the wizard's scope, creating a surface where malicious instructions in these files could attempt to influence the agent's output. \n
- Ingestion points: Project README, .env files, and GitHub workflows as specified in SKILL.md. \n
- Boundary markers: The agent is instructed to present the proposed list of stages for human confirmation before generating the script. \n
- Capability inventory: The agent can write shell scripts and use local validation tools like bash and shellcheck. \n
- Sanitization: The skill relies on human-in-the-loop review and the fact that the agent does not execute the generated script itself. \n- [COMMAND_EXECUTION]: The agent uses standard utilities (bash -n, shellcheck, chmod) to verify the integrity and permissions of the generated shell script.
Audit Metadata