The Agent Skills Directory

[Command Execution] (SAFE): The skill utilizes standard Node.js development commands such as npm test, npm run build, and npm ci. These operations are well-aligned with the skill's primary purpose of verifying project integrity.
[Indirect Prompt Injection] (SAFE): The skill monitors command outputs to confirm success. 1. Ingestion points: Command output and test-results/ files. 2. Boundary markers: Absent. 3. Capability inventory: Execution of project-defined npm scripts. 4. Sanitization: Absent. This is considered safe as it is essential for the skill's verification logic.
[Prompt Injection] (SAFE): The skill employs strong instructional directives (e.g., 'Iron Law') to ensure the agent performs verification before claiming completion. These instructions serve to define a reliable task persona rather than to bypass safety guardrails.

starwards-verification