The Agent Skills Directory

[SAFE]: The skill provides structured templates for implementation planning. The inclusion of standard CLI commands like pytest and git within markdown examples is appropriate for a development-oriented agent and represents routine developer workflow functionality.
[PROMPT_INJECTION]: The skill processes external requirements to generate plans, which is a surface for indirect prompt injection. However, the enforced TDD structure and rigid task-based format limit the impact of adversarial input.
Ingestion points: External specifications or requirements provided to the agent (SKILL.md).
Boundary markers: The skill enforces a strict markdown header and task structure to delimit generated plans.
Capability inventory: The skill is restricted to writing plan documents to the docs/plans/ directory.
Sanitization: Relies on structural formatting via the provided implementation template rather than explicit content filtering.

writing-plans