The Agent Skills Directory

[SAFE]: The skill performs legitimate repository maintenance tasks using a restricted set of tools (Read, Glob, Grep, Write). It follows the best practice of requesting user confirmation via AskUserQuestion before performing any file system modifications.- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests and processes untrusted data from code comments and file metadata. * Ingestion points: Codebase files scanned via Glob and Grep in SKILL.md. * Boundary markers: None explicitly defined for scanned content. * Capability inventory: File read/write access and user interaction. * Sanitization: None mentioned; however, the process includes a human-in-the-loop step where the agent must ask for permission before writing the results to the register file, which serves as a mitigation.

tech-debt