The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is instructed to 'Run relevant verification commands from the task document when safe.' This functionality allows the agent to execute arbitrary shell commands defined within markdown files located in the docs/task/ directory.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by reading and implementing steps from external task documents. Malicious instructions within these files could attempt to manipulate the agent's behavior.
Ingestion points: docs/task/{ID}-{task-name}.md, TASKS.md, and AGENTS.md (via reading rules and workflow steps).
Boundary markers: The instructions do not specify any delimiters or ignore-instructions for the external data, relying instead on manual oversight.
Capability inventory: The skill is capable of file modification ('Apply the implementation steps') and executing shell commands ('verification commands').
Sanitization: The workflow requires a manual verification step ('Verify the task has been reviewed and approved for implementation') which acts as the critical security filter for third-party or untrusted task content.

implement