game-experience-review
Skill1 starsChanged 4 days agoMITView on GitHub
Review game experience through independent, code-informed player and critic perspectives, plan focused playtests, or compare builds with testable improvements.
In this repository25 other files it gives its agents.
- AGENTS.mdAGENTS.md
- image-quality-controlSkill
- reference-image-promptSkill
- suno-createSkill
- suno-music-explorerSkill
- suno-studio-downloadSkill
- consensus-change-reviewSkill
- consensus-review-loopSkill
- delegated-change-reviewSkill
- docs-context-curatorSkill
- godot-silent-reviewSkill
- implement-in-stagesSkill
- improve-code-commentsSkill
- review-gated-implementationSkill
- review-loopSkill
- review-testsSkill
- codex-session-recoverySkill
- codex-thread-reportSkill
- consensus-gated-grillingSkill
- grillingSkill
- opencode-session-toolkitSkill
- review-gated-grillingSkill
- sanitize-artifactsSkill
- skill-manual-onlySkill
- wait-whatSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 13
Follow the requested task.
- Distinctive: line 31
Use three complementary, read-only reviewers by default.
- Distinctive: line 44
Only for a requested blind exercise, restrict access and record expectations before revealing code, solutions, or later outcomes.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Game Experience Review
Review what players perceive, decide, do, and receive in return. Recommend improvements; implement only within the user's authorized scope.
Choose the task
Follow the requested task; default to review. Combine tasks only when requested.
| Task | Result |
review | Assess a game or representative section and recommend supported improvements. |
plan | Design a focused playtest; use existing material to frame questions, without adding an unsolicited game critique. |
compare | Compare named builds at matched moments and conditions; report unmatched coverage separately. |
Resolve the target, audience, intended experience, and question from the request and current project guidance. Record the source/build or working-tree state, evidence versions, and relevant settings; do not relabel old evidence as current.
Inspect relevant evidence
Every lens may read relevant code, content, tests, and documentation. Use material entry points and a task scope, not a fixed file whitelist; follow related implementation and evidence as needed. Trace player-visible information, available actions, rules, feedback, and consequences. Distinguish what the reviewer learns from source from what a player can know at that moment. Reading code supports informed analysis, not a claim of an unspoiled first encounter.
Reuse authorized project capture/replay tools and their dependency, window, save-isolation, and output rules. Missing support limits claims; it does not authorize building adapters or installing dependencies. Inspect actual material, not just passing assertions. State whether a claim comes from source, rendered frames, recorded motion, interaction, audio, or human sessions, and limit it accordingly. Code and static images cannot establish control feel; scripted success cannot establish natural discovery; agent behavior is not human response.
Independent perspectives
Use three complementary, read-only reviewers by default; honor a requested count or focus. Usually choose a player perspective, a critic, and the most relevant specialist.
| Lens | Focus |
| First-time player perspective | What a newcomer can infer from presented goals, clues, actions, and recovery. |
| Genre player | Whether the core activity offers interesting decisions, variation, and reasons to continue. |
| Critic | How theme, mechanics, characters, pacing, and presentation create a distinctive experience worth preserving. |
| Narrative and systems | Motivation, knowledge boundaries, causality, costs, agency, and fair consequences. |
| Interaction and accessibility | Legibility, input, feedback, and recovery under the specified settings. |
| Project-specific | The central experience, such as supervisory judgment or mixed-experience co-op participation. |
Read references/reviewer-packet.md, fill its fields, and put the complete brief in each dispatch prompt; do not rely on skill inheritance. Start fresh contexts (fork_turns: "none" where supported), without parent history, proposed findings, or other reviewer reports. Role names alone do not establish independence. If isolation or a reviewer is unavailable, disclose incomplete independent coverage.
Only for a requested blind exercise, restrict access and record expectations before revealing code, solutions, or later outcomes. If this cannot be maintained, label the result retrospective analysis. Normal player-perspective review remains code-informed.
Synthesize
Verify cited observations and investigate causes rather than automatically adopting a player's proposed fix. Merge common causes; preserve disagreement and effective choices. Separate observations from interpretations, risks, preferences, and hypotheses. Prioritize experience impact and evidence confidence separately; several agents are not several player samples. Do not force findings, resolve taste by voting, or treat deliberate uncertainty as a defect.
Return a concise result in the user's language, with scope, versions, perspectives, evidence, and limits:
- Review: Preserve effective moments; explain supported issues with locators, observation versus interpretation, impact/confidence, smallest useful change, tradeoff, and retest. No supported issue is a valid result.
- Plan: Give participants and prior knowledge, tasks, success/failure behavior signals, neutral debrief questions, and outcomes that would change the decision.
- Compare: Explain supported differences and unmatched coverage, preserving the same evidence limits for both builds.
End with up to three next actions and unresolved questions. Propose focused human checks where needed; do not fabricate reactions or generalize a short review to the whole game.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

