The Agent Skills Directory

[SAFE] (SAFE): The skill is purely instructional, providing behavioral guidelines and output templates for the AI.
[DATA_EXPOSURE] (SAFE): No sensitive credentials, hardcoded secrets, or suspicious file paths are present. The use of Read, Glob, and Grep tools is consistent with the stated purpose of analyzing and managing code projects.
[REMOTE_CODE_EXECUTION] (SAFE): No package installations (npm/pip) or remote script downloads (curl/wget) are defined.
[PROMPT_INJECTION] (LOW): The skill includes behavioral overrides and manual mode triggers (e.g., /brainstorm). While these modify AI behavior, they are within the intended operational scope and do not attempt to bypass core safety or ethical guardrails.
[INDIRECT_PROMPT_INJECTION] (LOW):
Ingestion points: Uses Read, Glob, and Grep to ingest file content.
Boundary markers: None explicitly defined in the prompts.
Capability inventory: Capable of writing and modifying code in IMPLEMENT mode.
Sanitization: Not present. While the skill reads untrusted data, this is a standard risk for development agents and not a malicious feature of this specific skill.

behavioral-modes