The Agent Skills Directory

[SAFE]: The skill is a standard code analysis tool that operates locally on the user's provided repository path. It does not perform unexpected network operations or access sensitive system files outside the repository scope.
[COMMAND_EXECUTION]: The skill executes local shell scripts (scripts/detect-dotnet-version.sh, scripts/collect-diff.sh, scripts/run-checks.sh) to perform its functions. These scripts are bundled with the skill and focus on git operations and .NET CLI commands (dotnet build, dotnet format, dotnet test). Input parameters such as --repo-root are validated to ensure they are directories or git repositories before execution.
[DATA_EXPOSURE_AND_EXFILTRATION]: No data exfiltration patterns were detected. The skill generates a report file locally in the docs/reviews/ directory. It explicitly excludes minified JavaScript and library files from its diff collection to reduce noise and potential exposure of non-source data. It contains an explicit rule to never include secrets in logs or reports.
[PROMPT_INJECTION]: The skill includes clear instructions to ignore generic 'review my code' requests and only activate on explicit invocation. This prevents hijacking of other language reviewers. It does not contain any patterns intended to bypass AI safety guardrails.
[DYNAMIC_EXECUTION]: While the skill uses python3 for JSON escaping of diff content, this is a safe application of a standard utility for data formatting rather than dynamic code generation or execution of untrusted input.

dotnet-reviewer