The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates project management workflows by executing GitHub CLI (gh) commands to interact with tracker issues. This includes gh issue list to search for existing documents, gh issue view to read content, and gh issue create or gh issue edit to store the drafted PRDs. These operations are restricted to the repository tracker and are part of the skill's core functionality for managing requirements.- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8) through its integration with external data sources. * Ingestion points: External data is ingested from GitHub issue bodies via gh issue view in the 'plan' and 'update' workflows. * Boundary markers: None identified. The skill relies on the structured PRD template to guide the model's focus. * Capability inventory: Subprocess calls include gh issue create and gh issue edit for modifying remote tracker state. * Sanitization: Includes a 'Quality Gates' checklist that validates PRD structure and content integrity (e.g., ensuring no placeholders, presence of measurable goals) before handoff to planning agents. The security risk is significantly mitigated by the disable-model-invocation: true flag in the skill frontmatter, which prevents the agent from triggering further tool executions while the skill is active.

writing-prds