The Agent Skills Directory

[SAFE]: The skill functions as a coordination layer for software development workflows. No malicious patterns or security risks were identified.
[COMMAND_EXECUTION]: The skill references the use of a local command-line tool harness to perform plan reviews (harness run plan-review). This appears to be an expected tool for the intended environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process repository-based artifacts like briefs and implementation plans, which could theoretically contain malicious instructions. Ingestion points: Artifacts in dev/briefs/, dev/plans/, and dev/issues/, along with user-provided requirements. Boundary markers: The workflow incorporates validation steps (review-spec, plan-review) that check artifacts against the repository's intent source before execution. Capability inventory: The skill acts as a coordinator, invoking other skills and the harness utility. Sanitization: The skill relies on the agent's interpretation and routing logic to validate content.

planning-workflow