The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from user-supplied markdown files, such as specifications and brainstorming notes, to generate task instructions and implementation hints. Ingestion points: Reads docs/specs/[id]/YYYY-MM-DD--feature-name.md, user-request.md, and brainstorming-notes.md from the project directory. Boundary markers: Employs a requirement taxonomy ([IMP], [SEF], [EXT]) and a 'Spec Fidelity Gate' to structure the ingestion process. Capability inventory: Writes task files (TASK-XXX.md), data models, and contracts to the filesystem; it also generates implementation commands for subsequent agent use. Sanitization: Performs a pre-check for vague language and requires a specific taxonomy for implementable requirements, though it does not provide formal escaping for content interpolation.
[COMMAND_EXECUTION]: The skill is instructed to run local analysis tools and reference other skills in the suite. Evidence: The instructions command the agent to run /graphify <project-root> to generate codebase insights. Evidence: The generated task files include specific commands to invoke the /skill:specs-kit-task-implementation skill for future steps.

specs-kit-spec-to-tasks