Conduct an in-depth requirements interview to flesh out a plan or spec file. Ask non-obvious questions about technical implementation, UI & UX, tradeoffs, and concerns — then write the finalized spec back to the file. Use when you need to flesh out requirements, refine a spec, clarify a plan, or gather details before implementation. Keywords: interview, spec, plan, requirements, flesh out, questions, clarify spec, requirements gathering.
Install
npx skillscat add wcygan/dotfiles/interview Install via the SkillsCat registry.
Interview: Spec & Plan Refinement
Conduct an in-depth interview to flesh out a plan or specification, then write the finalized spec back to the file.
Workflow
1. Load Context
If $ARGUMENTS references a file, read it to understand the current plan:
Read $ARGUMENTSIf no argument was provided, ask the user what topic or file they want to flesh out before proceeding.
2. Interview
Using the AskUserQuestion tool, conduct a thorough, non-obvious interview. Cover angles such as:
- Technical implementation: architecture decisions, data models, API design, performance requirements, edge cases, failure modes
- UI & UX: user flows, interaction patterns, error states, onboarding, accessibility
- Tradeoffs: build vs. buy, consistency vs. flexibility, simplicity vs. power, short-term vs. long-term
- Concerns: security, scalability, maintainability, cost, operational burden
- Constraints: timelines, team capabilities, external dependencies, non-negotiables
- Success criteria: how do we know it's done? what does "working" look like? what's out of scope?
Rules:
- Ask questions that expose hidden assumptions — not obvious ones
- Continue interviewing until the spec is complete; don't stop after a single round
- Each subsequent round should go deeper or cover an uncovered angle
- Surface contradictions and ambiguities in the existing plan and resolve them
3. Write the Spec
Once the interview is complete, write the finalized, detailed spec back to $ARGUMENTS (or ask for a filename if no argument was given).
Incorporate all decisions made during the interview. The output should be a complete, actionable spec — not just a summary of answers.