keep-on-going
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill identifies and executes shell commands specified within task data files (referred to as beans). Instructions for this behavior are found in the Runnable gates section of SKILL.md and the references/approvals.md file, which define a process for running commands found in the
- Run X and expect Y pattern.
- [INDIRECT_PROMPT_INJECTION]: The skill processes task definitions that could potentially harbor malicious instructions if the data source is compromised. 1. Ingestion points: Markdown files within the .beans/ directory. 2. Boundary markers: Structured headers like ## Proof of Work and specific patterns such as EXPECT: are used to delimit data. 3. Capability inventory: The skill utilizes shell command execution and the beans CLI tool. 4. Sanitization: A detailed security policy in references/approvals.md requires the agent to audit all commands for risks such as credential harvesting, destructive file operations, or unauthorized network activity before approval.
Audit Metadata