| name | hybrid-prototype |
| description | Fast-lane prototype skill for the hybrid workflow. Builds a playable prototype in 2-3 days with minimal process overhead. Designed for discovery phase. |
| argument-hint | [concept-description] |
| user-invocable | true |
| allowed-tools | Read, Glob, Grep, Write, Edit, Bash, Task |
| agent | prototyper |
| isolation | worktree |
Overview
This skill implements the Discovery Phase fast lane as described in docs/framework/hybrid-workflow.md. It is intentionally lightweight: no formal GDD, no architecture, no epic breakdown. Just build it, play it, decide.
Time budget: 1-3 days.
Agents involved: creative-director, game-designer, prototyper, godot-specialist (or engine equivalent).
Phase 1: Concept & Question (5 minutes)
Read the concept description from the argument. State the one core question this prototype must answer. If the concept is vague, ask the user to clarify before proceeding.
Examples of good questions:
- "Does the combat feel responsive with 200ms input lag?"
- "Is resource scarcity actually fun, or just frustrating?"
- "Does the movement mechanic support the intended platforming challenges?"
Bad question: "Is this game fun?" (Too broad. Narrow it down.)
Ask the user: "The core question for this prototype is: [question]. Proceed?"
Phase 2: Plan (15 minutes)
Define the minimum viable prototype in 3-5 bullet points:
- What is the absolute minimum code to answer the question?
- What can be hardcoded / placeholder / skipped?
- What is the success criteria? (e.g., "Player can complete 3 jumps in a row without dying")
Present the plan to the user and ask for confirmation.
Phase 3: Build (1-2 days)
Ask: "May I create the prototype directory at prototypes/[concept-name]/ and begin implementation?"
If yes, create the directory. Every file must begin with:
// PROTOTYPE - NOT FOR PRODUCTION
// Question: [Core question being tested]
// Date: [Current date]
Rules for prototype code:
- Hardcode values freely
- Use placeholder assets (colored squares, simple shapes)
- Skip error handling
- Use the simplest approach that works
- Copy code rather than importing from production
- NEVER import from
src/ — prototypes are isolated
Run the prototype as you build. Test continuously. Fix blockers, but don't polish.
Phase 4: Playtest (2-4 hours)
Play the prototype yourself. Then ask the user to play it. Collect observations:
- What worked?
- What felt bad?
- Did it answer the core question?
- Any surprising discoveries?
Document findings informally — a bulleted list is fine.
Phase 5: Decide (30 minutes)
Collaborate with creative-director and game-designer (via Task or conversation) to make a decision:
| Verdict | Meaning | Next Step |
|---|
| ITERATE | Core is promising, but needs adjustment | Run /hybrid-prototype [revised-concept] |
| PIVOT | The concept doesn't work, but a related one might | Run /concept-brainstorm or /hybrid-prototype [new-direction] |
| PRODUCTIONIZE | It's fun and proven — move to production | Begin GDD in /design-system, architecture in /create-architecture |
| KILL | It's not fun and no clear fix | Stop. The prototype report is the deliverable. |
Update prototypes/[concept-name]/DECISION.md with:
# Prototype Decision: [Concept Name]
## Question
[Core question]
## Result
[What happened]
## Verdict
[ITERATE / PIVOT / PRODUCTIONIZE / KILL]
## Reasoning
[Why]
## Next Steps
[What to do next]
Ask: "May I write the decision to prototypes/[concept-name]/DECISION.md?"
Phase 6: Done
Output a summary to the user: the core question, the verdict, and the next step.
If PRODUCTIONIZE: remind them to switch to the Production phase workflow (/design-system, /create-architecture, etc.)
If ITERATE / PIVOT / KILL: no further action needed.
Constraints
- Prototype code must NEVER import from production source files
- Production code must NEVER import from prototype directories
- If productionizing, rewrite from scratch — do not refactor prototype code
- Timebox strictly: if it's not working after 3 days, kill or pivot
- Keep the question narrow — one prototype, one question
- Workflow isolation: This skill explicitly bypasses
production/review-mode.txt. Any stale review-mode state from a previous full OCGS session is ignored — the hybrid fast lane always runs without formal gates.
Differences from Full /prototype Skill
| Aspect | /prototype (Full OCGS) | /hybrid-prototype (Fast Lane) |
|---|
| Review mode gates | Solo / Lean / Full | None (always fast) |
| Creative Director review | Formal gate spawn | Informal chat/Task |
| Report format | Formal REPORT.md | Lightweight DECISION.md |
| Agents involved | All tiers | 4 core roles only |
| Time to verdict | 1-3 days + review overhead | 1-3 days total |
| Next step on PROCEED | Formal GDD + ADR | Start GDD when ready |
