The Agent Skills Directory

[SAFE]: The skill uses established .NET tools such as dotnet build and dotnet list package --vulnerable to gather metrics. These are standard developer operations.
[SAFE]: The instructions emphasize using high-confidence findings for grading and warn against using raw detector counts without triage, promoting accuracy and preventing false positives.
[SAFE]: Security posture evaluation includes checking for vulnerable packages and hardcoded secrets using standard patterns, which is a best practice.
[SAFE]: The skill operates on local project files within the agent's provided environment and does not exhibit any network exfiltration, obfuscation, or persistence mechanisms.
[SAFE]: The rubric-based assessment process is transparent and relies on specified MCP tools (detect_antipatterns, get_project_graph, etc.) for data collection without executing arbitrary remote code.

health-check