The Agent Skills Directory

[SAFE]: The skill is restricted to read-only operations (Read, Grep, Glob) and performs validation against internal documentation. No signs of data exfiltration, remote code execution, or privilege escalation were found.
[PROMPT_INJECTION]: The skill processes untrusted JSON data from grid files, which represents an indirect prompt injection surface. Ingestion points: Reads a single JSON file (body.json). Boundary markers: Absent. Capability inventory: Uses Read, Grep, and Glob via the Task tool. Sanitization: Absent. The risk is considered safe given the restricted toolset and the sub-agent's narrow scope.

grid-validator