The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by accepting user-supplied test names or cargo arguments which are then interpolated into shell commands. Ingestion points: User arguments provided to the /freenet:linux-test command. Boundary markers: None present. Capability inventory: Execution of docker build and a local runner script docker/test-runner/run.sh with user arguments. Sanitization: None; the instructions explicitly direct the agent to pass cargo arguments through directly.
[DYNAMIC_EXECUTION]: The skill dynamically constructs shell commands by interpolating user-provided input into the execution of docker/test-runner/run.sh. Evidence: The instructions in Step 3 specify running docker/test-runner/run.sh <command args> where arguments are user-supplied. This pattern presents a potential command injection surface if the agent tool does not perform its own validation, although this behavior is core to the skill's utility as a flexible test runner.

linux-test