The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to download the pvanalyze tool from its official GitHub repository (github.com/adityamandaleeka/pvanalyze) or via the standard .NET tool registry. These are well-known services and the sources are consistent with the tool's intended diagnostic purpose.
[COMMAND_EXECUTION]: The skill uses standard .NET command-line interface utilities such as dotnet tool install, git clone, and dotnet build to set up the environment. It further executes pvanalyze and dotnet-trace to perform data analysis and trace collection, which are routine operations for software performance engineering.
[INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting and processing .nettrace files, which are external artifacts provided by the user or generated by tracing tools.
Ingestion points: Specified in SKILL.md (Workflow steps 1 and 2), primarily through path arguments to the pvanalyze command.
Boundary markers: None explicitly defined in the prompts to delimit the contents of the trace file from instructions.
Capability inventory: The skill performs subprocess calls to pvanalyze and dotnet-trace across several files, including SKILL.md and references/commands.md.
Sanitization: No specific sanitization or validation of the binary trace file structure is mentioned, though the tool itself (pvanalyze) is responsible for safe parsing of the diagnostic data.

pvanalyze