The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes high-capability diagnostic tools, such as dotnet-trace, dotnet-dump, and dotnet-gcdump, to capture runtime data and process state. This includes the ability to collect full memory dumps, which inherently contain all data present in the process memory, including potential secrets, credentials, or personally identifiable information (PII).
[PRIVILEGE_ESCALATION]: The documentation indicates that executing diagnostic tools may require elevated privileges, such as running as root on Linux or macOS, to access process information and performance counters.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for installing official .NET diagnostics tools via the dotnet tool install command or from official Microsoft download sources when the tools are missing from the environment.
[INDIRECT_PROMPT_INJECTION]: The skill involves the agent processing and analyzing output from diagnostic tools, such as stack traces, exception reports, and heap statistics. If an external attacker can influence the profiled application's state (e.g., through user-controlled input that appears in exception messages or thread names), they could potentially introduce instructions intended to influence the agent's reasoning.
Ingestion points: Textual output from commands like dotnet-trace report, dotnet-counters monitor, and dotnet-dump analyze (references/commands.md).
Boundary markers: None specified; the agent is instructed to read the tool output directly.
Capability inventory: The skill can execute shell commands, perform file operations, and install global .NET tools.
Sanitization: No explicit sanitization or filtering of the diagnostic tool output is defined before the agent performs its analysis.

profiling