The Agent Skills Directory

[SAFE]: No malicious patterns, hidden code, or unauthorized network operations were detected. The skill is purely informational and template-driven.
[DATA_EXPOSURE]: The skill reads project-level configuration files such as CONTRIBUTING.md and .github/ISSUE_TEMPLATE/ to adapt its templates to the project's language and standards. This is a standard localized operation with no evidence of exfiltration.
[PROMPT_INJECTION]: An indirect prompt injection surface exists via the feature description input in SKILL.md. Evidence chain: 1. Ingestion point: User-supplied argument for the /requirement command. 2. Boundary markers: None explicitly defined in the instructions. 3. Capability inventory: Limited to Read, Write, and Grep tools. 4. Sanitization: Not explicitly present. Despite the surface, the risk is negligible as the skill only uses this data for documentation generation and does not perform high-risk operations like shell execution.

requirement