The Agent Skills Directory

Indirect Prompt Injection (HIGH): The skill defines parameters profile_type and duration which are used to construct shell commands for Go profiling.
Ingestion points: Parameters profile_type and duration defined in SKILL.md.
Boundary markers: Absent. There are no delimiters or specific instructions to treat these values as untrusted data or to separate them from the command structure.
Capability inventory: The skill documentation provides command templates for go tool pprof and curl -o which involve shell execution and file writing.
Sanitization: Absent. The skill does not provide instructions to the agent to validate or escape these parameters, creating a vulnerability where an attacker could provide values like cpu; rm -rf / to execute arbitrary commands.
Command Execution (MEDIUM): The skill's primary function is to guide the agent in executing various shell commands (go tool, go test, curl). While standard for a profiling tool, these capabilities serve as a high-impact primitive for exploitation if the agent's logic is subverted through the identified injection surface.

go-performance