The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches its code quality standards from the author's official GitHub repository (github.com/somnio-software/somnio-ai-tools). These downloads consist of markdown-based rule files used to configure the auditor's logic during a run.
[COMMAND_EXECUTION]: To provide accurate metrics on code complexity and coupling (CA1502/CA1506), the skill uses the Bash tool to execute dotnet build with specific diagnostic flags enabled. This is a standard and necessary operation for the skill's primary function as a code quality auditor.
[DYNAMIC_EXECUTION]: The skill invokes the .NET compiler and analyzers at runtime to evaluate the target codebase. This dynamic analysis is limited to the local environment and the project provided by the user, representing normal behavior for development tooling.
[INDIRECT_PROMPT_INJECTION]: As an auditing tool, the skill must process untrusted source code from external repositories. While it contains no inherent vulnerabilities, its access to tools like Bash and Write necessitates caution when reading files that could contain malicious instructions.
Ingestion points: C# source files (.cs), project files (.csproj), and other configuration files within the audited repository.
Boundary markers: The skill employs detailed reference instructions (references/*.md) and a mandatory structured report template (assets/report-template.md) to restrict agent output and prevent the execution of embedded instructions.
Capability inventory: The skill is granted access to the Read, Write, Bash, Agent, and WebFetch tools.
Sanitization: The skill uses a tiered sub-agent architecture to isolate different analysis tasks, ensuring that the final synthesis remains focused on technical evidence rather than any instructions encountered within the data.

dotnet-best-practices