The Agent Skills Directory

[SAFE]: The skill is a procedural tool for software development and does not exhibit any malicious behavior. It includes explicit 'Scope Discipline' instructions that prevent the agent from modifying or cleaning up code adjacent to the task, which reduces the risk of unintended side effects or unauthorized changes. The use of an implement-test-verify cycle ensures that all changes are validated before being committed.\n- [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes external task specifications. This is a common characteristic of development-focused skills and is addressed here by a structured execution discipline. \n
Ingestion points: Processes tasks or specifications from external files like task_plan.md, docs/specs/, or direct user input. \n
Boundary markers: The skill does not explicitly define delimiters or boundary markers for the ingested task data. \n
Capability inventory: The skill utilizes Agent tool calls for subagent dispatch, shell execution for testing (e.g., npm test), and extensive file-system modification capabilities. \n
Sanitization: There are no instructions for sanitizing or escaping the content of external task specifications.

incremental-impl