| name | refine-story |
| description | Interactive story clarification for the aiagent project. The story-refiner agent explores the codebase, asks focused question batches, and iterates until all unknowns are resolved. Outputs a refined story with acceptance criteria and a Relevant Codebase Context section. |
| user-invocable | true |
| argument-hint | add a new tool that does X |
| context | fork |
Story Refinement
Refine the following story:
Instructions
You are running an interactive Story Refinement session using the story-refiner agent.
Step 1: Codebase Exploration
Use the Agent tool with subagent_type: "story-refiner" and pass it the story description. Ask it to:
- Explore the relevant parts of the aiagent codebase
- Generate a focused batch of 3–6 clarifying questions — not open-ended brainstorming, specific questions with clear answers
Step 2: Present Questions
Present the questions to the user and collect answers.
Step 3: Iterate
Pass the answers back to the story-refiner agent. Ask it to:
- Either generate a follow-up batch of questions (if unknowns remain)
- Or produce the final Refined Story and Relevant Codebase Context output
Step 4: Output
When the story-refiner declares the story complete, present the full output to the user:
- Refined Story (Goal, Acceptance Criteria, Scope Boundaries, Known Constraints, Resolved Unknowns, Definition of Done)
- Relevant Codebase Context (table of files/services/patterns the architect and developer need to read)
Important
- The story-refiner does NOT write code — it only asks questions and synthesizes answers
- This skill is interactive — loop with the user until the story is complete
- The output of this skill is the input to
/design-review