The Agent Skills Directory

[COMMAND_EXECUTION]: The skill incorporates standard shell commands for code verification and quality assurance.
Evidence: SKILL.md requires the use of npm test, npm run build, npm run lint, and tsc --noEmit as part of the verification checkpoints after each batch of tasks.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process implementation plans, which represent a surface for external data ingestion.
Ingestion points: The skill ingests tasks and plans defined in user-provided implementation documents.
Boundary markers: The skill uses a structured batching rhythm (3 tasks per batch) and mandatory checkpoints to provide natural process boundaries.
Capability inventory: Shell execution capabilities are limited to standard development tooling (npm, tsc).
Sanitization: The skill emphasizes security checks like parameterizing inputs and avoiding hardcoded secrets within the implemented code.

plan-execution