The Agent Skills Directory

[Command Execution]: The skill uses local shell commands including git merge-base, git diff, and git ls-files to identify changed code. These are standard development operations used here to scope the analysis to the contributor's specific changes.
[Processing Untrusted Data]: The skill ingests C# source code and diff hunks to perform pattern matching against established coding rules. As the skill processes user-generated content, there is a theoretical surface for indirect prompt injection where instructions could be embedded in code comments. However, the skill is structured to use these inputs strictly as data for analysis.
Ingestion points: Diff hunks from git diff and file contents via view (SKILL.md).
Boundary markers: None explicitly defined for the AI's consumption of code snippets.
Capability inventory: File reading (view), git operations (git), and targeted file modification if the user explicitly requests a fix.
Sanitization: The skill relies on pattern matching for specific C# rules rather than executing the provided code.

code-quality