Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add kyh/vibedgames --skill design-lensesgit clone --depth 1 https://github.com/kyh/vibedgamesWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/kyh/vibedgames/design-lenses)<a href="https://agentmods.dev/skills/kyh/vibedgames/design-lenses"><img src="https://agentmods.dev/badge/skills/kyh/vibedgames/design-lenses/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/kyh/vibedgames/design-lenses"><img src="https://agentmods.dev/badge/skills/kyh/vibedgames/design-lenses.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00038 | $0.02275 |
| Opus 5.5 | $0.00015 | $0.00910 |
| Sonnet 5.5 | $0.00008 | $0.00455 |
| Haiku 4.5 | $0.00004 | $0.00228 |
Grade A, and why
design-lenses scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 175 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Design lens review
A lens is a viewpoint plus questions, not a rule (Schell). This skill runs a battery of lenses against a game and outputs findings an agent can implement. The defining failure of prompt-built games: mechanically correct, sensorially and emotionally dead. The lenses find where.
Procedure
- Play first, cold. Load like a first-time visitor. Record: time to first input, time to first win/lose feedback, what you understood without instructions, where you got bored. Play twice — once naive, once trying to break it (spam inputs, resize, lose on purpose, idle).
- Write the MDA sentence: "This game wants the player to feel X via dynamic Y produced by mechanics Z." Can't write it? That's finding #1 — no essential experience. Then write the loop contract: "player does [verb] to achieve [objective] while [pressure] creates risk; success → [reward], failure → [cost/retry]". Any bracket you can't fill is a finding too.
- Run the rejection gates (below). Each failed gate is an automatic
blocker— no lens needed. - Pick 8–12 lenses. Always: Essential Experience, Fun, Feedback, Juiciness, Accessibility. Add by genre (Skill vs Chance for arcade, Curiosity for puzzle, the Toy for sandbox/physics).
- Answer with evidence — a timestamp, a code path, a missing sound — never vibes.
- Severity:
blocker(player quits or never understands),major(understood but flat),minor(polish),idea(opportunity). - Findings as lens → evidence → fix, fixes concrete ("80ms hit-stop + 4px shake on collision", not "juicier"). Order: blockers, then cheapest major wins (audio + shake + score popups = highest fun-per-line).
- Re-play after fixes, re-asking only the failed lenses and gates.
Rejection gates
Binary go/no-go tests, self-applied in active play before claiming a game is fun. Any "true" = automatic blocker; fix or iterate before polishing:
- The first 30 seconds lack a real decision.
- The player can ignore the main mechanic and still progress.
- The objective is unclear without reading source code or instructions.
- Failure arrives before the player can understand why.
- Challenge is only "more things", never better combinations.
- Rewards change nothing — not strategy, score, progression, or feel.
- The space is decorative and shapes no decisions.
- The game is fun only in the designer's explanation, not in active play.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago Changed · -80 tokens per session 1e4b809f90a9
- 1mo ago First seen · 175 lines · 118 tokens per session scan A 9efb959b9feb
design-lenses is a skill published in the GitHub repository kyh/vibedgames (61 stars, last pushed today), licensed MIT. It adds 38 tokens to every session and 2,275 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
review-system-design
Review a system design someone else provided - screenshot, drawio, Mermaid, slides, doc, or IaC - by extracting it into a confirmed fact sheet, then scoring it on the nine axes.
codebase-review
Review an entire codebase for architecture, engineering health, and exploitable risk; generate a prioritized remediation plan, an evidence-anchored system knowledge document, or both.
code-review
Run an AI-assisted PR code review using multi-layer lenses with confidence scoring.
review-ticket
Review a ticket or PR through focused specialist lenses: scope, architecture, security, tests, AC coverage, and PR metadata.
common-feedback-reporter
Pre-write audit for skill violations: checks planned code against loaded skill anti-patterns before any file write. Use when writing Flutter/Dart/TS code or editing SKILL.md files with active project skills. Load as composite; on auto-fixed violation, also load +common/common-learning-log.
common-code-review
Conduct high-quality, persona-driven code reviews. Use when reviewing PRs, critiquing code quality, or analyzing changes for team feedback.