acceptance-criteria

name	acceptance-criteria
description	Defines acceptance criteria from requirements through persona analysis and refinement. Use when clarifying requirements, defining what done looks like, or writing acceptance criteria.

Acceptance Criteria

STARTER_CHARACTER = 🎯

Define what success looks like before thinking about code. Stay in the problem space — if you're naming functions, data structures, or config keys, you've left it.

Process

1. Restate the requirement in your own words

2. Identify personas — who will actually interact with this feature, directly or indirectly? Don't settle for the obvious ones. Think about:

   • Who triggers it? (human, agent, CI, cron?)
   • Who consumes the output?
   • Who configures it?
   • Who debugs it when it breaks?

   Present the personas to the user before continuing. You are likely wrong or incomplete — get confirmation.

3. Walk through each persona's real workflow — not the idealized flow, the actual one:

   • What are they doing right before they encounter this feature?
   • What do they see/do step by step?
   • What tells them it worked?
   • What tells them it didn't?

   This is where weak personas get exposed. If you can't describe a concrete, realistic workflow, the persona is too vague.

4. From those workflows, identify:

   • Inputs: What data/events trigger this?
   • Outputs: What should change? What should users see?
   • Boundaries: What's explicitly out of scope?
   • Unknowns: What needs clarification?

5. If unknowns exist, ask the user — don't guess on important decisions

6. Write criteria to playground/{feature-name}-criteria.md:

   # [Feature Name] Acceptance Criteria
   
   ## Personas
   - [Persona]: [one-line description of their relationship to this feature]
   
   ## Must Have
   - [ ] Criterion 1 (specific, testable)
   - [ ] Criterion 2
   
   ## Out of Scope
   - Item explicitly excluded
   

Use a short, hyphenated feature name (e.g., user-auth, invoice-export) consistently across all playground files.

Gate

Invoke /refinement-loop on the criteria file with these checks:

• Every persona has at least one criterion that specifically serves them?
• Specific enough to test? (not "works well" but "returns X when given Y")
• Complete? Walk through each persona's workflow — any gaps?
• Bounded? Clear what's NOT included?

Do not consider this done until every criterion passes every check.