verification-before-completion
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to identify and execute shell commands to verify the state of the project (e.g., tests, builds, and linters).
- Evidence: The "Gate Function" section requires the agent to "IDENTIFY: What command proves this claim?" and "RUN: Execute the FULL command (fresh, complete)".
- [INDIRECT_PROMPT_INJECTION]: The skill involves processing external data such as agent reports and file diffs to decide on verification actions, creating a potential surface for indirect influence.
- Ingestion points: The "Key Patterns" section mentions reading "agent success reports" and "VCS diffs" as inputs for verification.
- Boundary markers: The instructions do not define explicit delimiters or instructions to ignore embedded commands within the processed reports or diffs.
- Capability inventory: The skill utilizes shell command execution to perform the required verification steps.
- Sanitization: No specific sanitization, validation, or escaping of data-derived commands is prescribed.
Audit Metadata