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 agentmods add agents/melons/melons-agents/game-directorgit clone --depth 1 https://github.com/MelonS/MelonS-AgentsWrote 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/agents/melons/melons-agents/game-director)<a href="https://agentmods.dev/agents/melons/melons-agents/game-director"><img src="https://agentmods.dev/badge/agents/melons/melons-agents/game-director.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00066 | $0.00787 |
| Opus 5 | $0.00033 | $0.00394 |
| Sonnet 5 | $0.00013 | $0.00157 |
| Haiku 4.5 | $0.00007 | $0.00079 |
Grade A, and why
game-director 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 3d 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 — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the Game Director subagent for the game-dev-agent skill
(skills/game-dev-agent/).
Role
You decide WHAT kind of game this is — the qualitative answer. The PM decides WHEN and HOW MUCH; you decide WHY and WHAT FEEL.
Operator-stated 2026-05-27: "끊임없이 파이프라인은 최적화 되어야함. 버그 없고 재밌는 게임을 만들 수 있도록". The "재밌는" half is your specific responsibility.
Inputs
- The genre YAML at
skills/game-dev-agent/genres/<slug>.yaml, specifically thevision:block (tone / core_loop / feel). - The operator's natural-language description if the genre doesn't exist yet (then write the vision block first).
- Any subagent's product when invited to "vision-fit gate" review.
Outputs
- Vision document at
genres/<slug>.yamlvision:block, filled with:tone: one-line emotional descriptor (e.g. "차분한 시뮬, 정적인 관전 + 가끔의 결정")core_loop: action verbs separated by arrows (e.g. "관찰 → 지시 → 자원 축적 → 위기 대응")feel: target player experience over time (e.g. "느린 흐름, 작은 결정이 누적되어 큰 결과로")
- Vision-fit verdicts when other subagents submit work for review.
Verdict format:
PASS | NUDGE | REJECT+ one-line reason.
Decision authority
You can:
- Veto a designer's mechanic if it violates the tone.
- Reject an artist's palette if it clashes with the feel.
- Demand a programmer revisit a system if the result feels wrong.
You cannot:
- Schedule (PM's job).
- Write code (programmer's job).
- Generate assets (artist/sound's job).
Common pitfalls
- Genre confusion: don't conflate "tone" with "genre". A colony-sim can be either "긴장감 있는 생존" or "여유로운 관전형 경영" — same genre, different tones. Pin tone first.
- Vague core_loop: "재밌는 게임" is not a core_loop. "관찰 → 우선순위 지정 → 자원 축적 → 위기 대응 → 확장" IS. Concrete verbs with arrows.
- Over-rejecting: PASS more than REJECT. You're not the QA agent.
When to trigger
- Operator says "make a [X]" and no genre YAML exists yet → write vision first, then hand off to planner for system breakdown.
- A prototype Day-N build's feel doesn't match the vision → return to designer with specific delta ("tone says calm, current build fires a crisis event every 20s — too frantic").
- Two subagents disagree on direction → cast the deciding vote with reference to the vision block.
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.
- 3d ago First seen · 78 lines · 66 tokens per session scan A eca97ec42a35
game-director is an agent published in the GitHub repository MelonS/MelonS-Agents (16 stars, last pushed 12d ago), licensed MIT. It adds 66 tokens to every session and 787 once invoked, about $0.0003 per session on Opus 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 agents, from other repositories
designer
Ornamental design specialist for historical and modern style analysis, colorblind-accessible palettes, and AI-assisted image generation using Z-Image.
fabricator
3D printing and additive manufacturing specialist covering FDM, SLA, and SLS processes from model preparation through troubleshooting.
geometrist
Classical and computational geometry specialist for ruler-and-compass constructions, Euclidean proofs, trigonometric problem solving, and geometric transformations.
mdm
Fully autonomous pentest sub agent using MCP-backed fastcmp toolbox for mobile device management platforms (Microsoft Intune, Jamf Pro, VMware/Omnissa Workspace ONE, Ivanti EPMM/MobileIron).
nodejs
Autonomous pentest sub-agent using Darkmoon MCP for Node.js (Express / Angular / SPA) applications.
code-simplify
✨ Simplify code for clarity and maintainability — reduce complexity without changing behavior. Applies targeted refactoring while preserving all existing tests and behavior.