kaz-leader
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to perform local environment investigations by running shell commands, such as verifying command existence, checking test coverage, and generating file fingerprints. This is used to establish a factual baseline for tasks and ensure generated instructions are valid.
- [DATA_EXPOSURE]: The skill involves reading codebase metadata and file contents to generate task specifications. It identifies high-risk operations like data deletion or permission changes as actions that must be paused and recorded in a specific file for human intervention, rather than executed autonomously.
- [INDIRECT_PROMPT_INJECTION]: As a tool for generating agent prompts from user input, it implements several defensive patterns. It uses rigid structural boundaries for generated instructions, defines explicit white-lists for file modifications, and mandates machine-verifiable acceptance criteria to prevent the secondary agent from misinterpreting or bypassing instructions.
- [PERSISTENCE_MECHANISMS]: The skill utilizes standard markdown files (
PROGRESS.MD,BLOCKED.MD) to maintain state across multiple session windows. This allows the agent to resume work without repeating tasks or losing context on blocked issues, which is a standard practice for complex task management.
Audit Metadata