# Gsp Playability Verifier

> Use when verifying a game build against its quality target before claiming it is complete.

- Skill: `mike007jd/gsp-playability-verifier-2` (Agent Skill, multi-file: 11 files)
- Install (CLI): `npx skillmds@latest add mike007jd/gsp-playability-verifier-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/mike007jd/gsp-playability-verifier-2/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: mike007jd (https://skillmd.com/u/mike007jd)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/mike007jd/gsp-playability-verifier-2

---


# Game Playability Verifier

## Goal
Verify the build against its **actual target**, not a generic low bar.

## Outputs

Follow the `gsp-orchestrator` output strategy:
- **inline** (default): present verification results in conversation.
- **minimal** or **full**: write `docs/game-studio/quality-report.md`.

## Use
- `./shared/templates/quality-report.md`
- `./shared/checklists/first-playable-rubric.md`
- `./shared/checklists/polished-prototype-rubric.md`
- `./shared/checklists/playability-verifier-output-checklist.md`
- `./shared/checklists/ui-ux-hard-rules.md`

## Quality-aware verification
- If the target is `first-playable`, use `./shared/checklists/first-playable-rubric.md`.
- If the target is `polished-prototype`, also use `./shared/checklists/polished-prototype-rubric.md`.
- If the target is `production-feature` or `live-patch`, verify the requested feature set plus regression risk in touched areas.

## Minimum checks
- boots
- reaches the correct actionable state
- requested verbs work
- requested critical features in scope are present
- HUD / affordances are readable
- playfield is not obviously sacrificed to UI clutter
- no blocking runtime errors
- the build quality matches the selected target

## Reporting rules
For each critical check, report:
- pass / fail / partial
- evidence source
- short note or repair direction

## Important
Do not declare success merely because something runs.
The verifier exists to catch “technically alive, product-wise weak” outcomes.
When runtime evidence is missing, say so explicitly.
In a multi-agent build flow, this skill is the default verifier stage after implementation review, not an optional final flourish.