The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to "launch the intended target and establish control and observation" to confirm method feasibility. It also requires the execution of commands for unit tests, CRAP index calculations, and other project gates based on the repository's existing configuration.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing external repository data and utilizing it to drive command execution.
Ingestion points: In SKILL.md, the agent is directed to "Inspect the current repository rather than relying on conversation alone" and to read design inputs, ADRs, and project standards.
Boundary markers: The instructions do not define explicit delimiters or "ignore embedded instructions" warnings for the content read from the repository.
Capability inventory: The agent possesses the capability to launch application targets and execute arbitrary shell commands defined in the project's executable configuration (tests, linting, and complexity tools).
Sanitization: There is no mention of sanitizing or validating the commands extracted from the project configuration before they are executed in the environment.

implementation-spec