The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands using pwsh and the dotnet CLI to perform repository inspection, project building, and benchmark execution.\n
Evidence: SKILL.md specifies commands such as pwsh -NoProfile -File "<skill-root>/scripts/check-benchmark-requirements.ps1" and dotnet run -c Release ....\n- [DYNAMIC_EXECUTION]: Bundled PowerShell scripts are used for repository state detection. These scripts invoke the dotnet CLI as a subprocess to verify environment details.\n
Evidence: The script scripts/check-benchmark-requirements.ps1 uses [System.Diagnostics.Process] to execute dotnet --version and capture its output.\n- [INDIRECT_PROMPT_INJECTION]: The skill analyzes repository source code and project metadata to design benchmark experiments. Malicious instructions placed in repository content (e.g., in source code comments) could potentially influence the agent's behavior during its analysis or code generation phases.\n
Ingestion points: The agent reads source implementation, call sites, tests, and project files (SKILL.md, Step 3).\n
Capability inventory: The agent can create/modify C# and project files and execute CLI tools.\n
Boundary markers: The skill does not define specific delimiters or instructions to prevent the agent from acting on instructions found within repository content.\n
Sanitization: No content sanitization or filtering mechanisms are specified for the ingested repository data.

dotnet-benchmark