The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides and executes a wide range of diagnostic commands using .NET CLI tools such as dotnet-counters, dotnet-trace, dotnet-dump, and dotnet-gcdump. These commands are used to attach to running processes, collect metrics, and generate memory dumps.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of diagnostic utilities via the dotnet tool install command, targeting official Microsoft package registries. It also references official documentation for direct downloads when global tools are not applicable.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it explicitly instructs the agent to read and follow the constraints and scope defined in a local AGENTS.md file within the target repository.
- Ingestion points: The AGENTS.md file is read at the beginning of the workflow (SKILL.md).
- Boundary markers: None; the agent is directed to confirm scope and constraints directly from the untrusted repository file.
- Capability inventory: The skill possesses capabilities for tool installation, process attachment, and full memory dump collection (references/commands.md).
- Sanitization: No sanitization or verification of the content within the AGENTS.md file is specified before the agent incorporates it into its execution plan.

profiling