# 09 Game Test And Player Evidence Plan

> Use when planning game QA, playtesting, balance, deterministic simulation, save migration, multiplayer network conditions, device performance, accessibility, localisation, security, certification, and release-candidate evidence.

- Skill: `peterbamuhigire/09-game-test-and-player-evidence-plan` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add peterbamuhigire/09-game-test-and-player-evidence-plan`
- Raw SKILL.md: https://api.skillmd.com/api/skills/peterbamuhigire/09-game-test-and-player-evidence-plan/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Web & Frontend
- Author: peterbamuhigire (https://skillmd.com/u/peterbamuhigire)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/peterbamuhigire/09-game-test-and-player-evidence-plan

---

# Game Test and Player Evidence Plan
<!-- dual-compat-start -->
## Use When
- A game milestone needs technical tests and observed-player evidence.
## Do Not Use When
- Programme testing policy is the task; use `01-test-strategy`.
- Do not substitute opinions for correctness or device evidence.
## Required Inputs
| Artefact | Source or provider | Required? | Behaviour when missing |
|---|---|---|---|
| Requirements, architecture, build manifest, targets and gates | Product, engineering, QA and release | Yes | Stop and issue a readiness gap. |
| Research, learning, culture, wellbeing, ads/privacy, network, accessibility, localisation and threat cases | Specialist owners | Conditional | Mark coverage not assessed. |
## Workflow
1. Build a risk matrix across unit, integration, simulation, content, device, network, security and access layers.
2. Specify seeds, clocks, tolerances, replay and save corruption/migration/interruption.
3. Test latency, loss, jitter, reorder, duplication, disconnect, crash and version mismatch online.
4. Profile release builds on named hardware for pacing, CPU/GPU, memory, loading, storage and thermal behaviour.
5. Separate correctness, comprehension, appeal, balance, learning, cultural acceptance, wellbeing/stopping, ad interruption/trust, commercial and release questions; no evidence class substitutes for another.
6. Test critical verbs with alternative input and redundant feedback.
7. Test every ad placement for expectation, disclosure, focus/input, decline/close, reward, cap/pace, save/recovery/stopping/exit, child treatment, accessibility, no-fill/offline/SDK failure and kill switch on real devices.
8. Stop on missing build identity, oracle, consent or accountable reviewer; recover with an evidence plan.
## Outputs
| Artefact | Consumer | Acceptance condition |
|---|---|---|
| Game test and player evidence plan | QA, design, engineering and release | Cases map to requirements, environments, oracles, evidence and decisions. |
## Evidence Produced
| Evidence | Reviewer | Acceptance condition |
|---|---|---|
| `GTEST-*` matrix | QA reviewer | Build, device, scene, seed, network and outcome are reproducible. |
## Capability and permission boundaries
Read and search are allowed; planning is read-only. Shared-service, player-data or security tests require authority and consent.
## Degraded mode
Return the narrowest qualified plan, mark unavailable checks `not assessed`, and retain blocked gates.
## Decision Rules
| Choice or condition | Action | Failure or risk avoided |
|---|---|---|
| Oracle absent | Redesign or classify exploratory. | False confidence. |
| Capture lacks identity | Reject release evidence. | Irreproducibility. |
| Core accessibility blocker | Block or approve explicit exclusion. | Menu-only claims. |
## Quality Standards
- Retain failed-path evidence; distinguish preference, comprehension, balance and correctness.
## Anti-Patterns
- Tautological tests. Fix: assert invariants.
- Editor-only testing. Fix: packaged builds.
- Average FPS only. Fix: frame pacing.
- Localhost multiplayer only. Fix: hostile profiles.
- Anecdotal playtests. Fix: question, cohort and disposition.
- Ad renders so it passes. Fix: test interruption, accidental tap, coercion, privacy, regret, trust and all failure paths.
## References
- [Test evidence matrix](references/game-test-evidence-matrix.md)
- [Playtest learning evidence](references/playtest-learning-evidence.md)
<!-- dual-compat-end -->