The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes a local Python script scripts/deferred_capture.py to store alternative project approaches. The command arguments (title and context) are dynamically populated from the brainstorming session data.
[SAFE]: The skill implements an automated review loop (Phase 6.5) using a subagent to verify the quality of the generated project brief. While this involves processing user-influenced content, it is a standard quality-assurance mechanism.
Ingestion points: The review loop in modules/spec-review-loop.md reads content from docs/project-brief.md.
Boundary markers: None identified in the review prompt template.
Capability inventory: The main agent applies text modifications to the specification document based on subagent feedback.
Sanitization: Not explicitly defined for the input passed to the review subagent.
[SAFE]: Integration with other skills like attune:war-room and attune:project-specification follows standard orchestration patterns within the vendor's ecosystem for project lifecycle management.

project-brainstorming