The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to analyze the project's history using git diff and git log. This is a standard operation for a development-focused assistant to understand code changes.
[DATA_EXPOSURE]: To provide accurate explanations, the skill reads commit history and conversation logs. The gathered information is processed locally to create documentation within the workspace's sub-directory, with no evidence of external transmission.
[PROMPT_INJECTION]: The skill processes untrusted input from git commits and previous user interactions. While this represents a surface for indirect prompt injection (Category 8), the risk is negligible as the skill only outputs local text files and explicitly restricts further model calls through its frontmatter configuration.

work-explainer