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 skills/yasuui/ystack/nano-banana-skillnpx skills add Yasuui/ystack --skill nano-banana-skillgit clone --depth 1 https://github.com/Yasuui/ystackWrote 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/yasuui/ystack/nano-banana-skill)<a href="https://agentmods.dev/skills/yasuui/ystack/nano-banana-skill"><img src="https://agentmods.dev/badge/skills/yasuui/ystack/nano-banana-skill.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 5.1 | $0.00071 | $0.02161 |
| Opus 5 | $0.00036 | $0.01081 |
| Sonnet 5 | $0.00014 | $0.00432 |
| Haiku 4.5 | $0.00007 | $0.00216 |
Grade A, and why
nano-banana-skill 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 6d 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 — 213 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Nano Banana Skill
The Artist Mindset
Nano Banana is not a prompt box. It is a collaborator. The quality of output is entirely determined by the quality of direction. Vague prompts produce vague images. Specific direction produces specific results.
Before generating any image, answer these four questions:
- What is the exact subject? (not "a developer" — "a terminal window on a dark glass desk, partially reflected, 3/4 angle from the left")
- What is the lighting source? (not "good lighting" — "single cool blue rim light from the top-left, ambient fill from below in warm amber, no direct frontal fill")
- What is the camera relationship to the subject? (close-up, wide establishing, isometric, top-down, eye level, slightly below)
- What mood or decade does this reference? (Apple keynote 2020, early Stripe.com, Vercel 2023, editorial NYT tech section)
If you cannot answer all four, you are not ready to generate.
Prompt Architecture
Structure every prompt in this order:
[Subject description] + [spatial positioning] + [lighting] + [surface/material] +
[background treatment] + [color palette] + [mood/atmosphere] + [technical specs] +
[negative prompt]
The negative prompt is as important as the positive
Always include a negative prompt. The default negative prompt for all brand assets:
--no stock photography, watermark, lens flare, excessive bokeh,
saturated gradients, cartoon, illustration style, human faces,
text overlaid, banner ads, corporate clipart, plugin icons,
emoji-style graphics, purple-to-blue AI gradient
Asset Types and Prompts
Substack Header Background
Goal: abstract, editorial, dark. The text sits on top — the background should not compete.
Prompt template:
Abstract architectural geometry, [specific material: brushed titanium / oxidized steel / dark concrete /
smoked glass] surface, extreme close-up macro photography, shallow focus, cool ambient light from
the left at 15 degrees, rich dark shadows, [accent color]-tinted atmospheric haze in the midground,
no horizon line, no recognisable objects, mood: late-night studio, editorial, considered
--ar 16:9 --no text, faces, plants, logos, purple gradients, stock look
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.
- 6d ago First seen · 213 lines · 71 tokens per session scan A a6cba04001c4
nano-banana-skill is a skill published in the GitHub repository Yasuui/ystack (1 stars, last pushed 5mo ago), licensed MIT. It adds 71 tokens to every session and 2,161 once invoked, about $0.0004 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 2026-08-31.
Other skills, from other repositories
openclaw-persona-forge
为 OpenClaw AI Agent 锻造完整的龙虾灵魂方案。根据用户偏好或随机抽卡， 输出身份定位、灵魂描述(SOUL.md)、角色化底线规则、名字和头像生图提示词。 如当前环境提供已审核的生图 skill，可自动生成统一风格头像图片。 当用户需要创建、设计或定制 OpenClaw 龙虾灵魂时使用。 不适用于：微调已有 SOUL.md、非 OpenClaw 平台的角色设计、纯工具型无性格 Agent。 触发词：龙虾灵魂、虾魂、OpenClaw 灵魂、养虾灵魂、龙虾角色、龙虾定位、 龙虾剧本杀角色、龙虾游戏角色、龙虾 NPC、龙虾性格、龙虾背景故事、 lobster soul、lobster…
manim-video
Build reusable Manim explainers for technical concepts, graphs, system diagrams, and product walkthroughs, then hand off to the wider ECC video stack if needed. Use when the user wants a clean animated explainer rather than a generic talking-head script.
env-scanner
Scan and audit Gemini CLI, Claude Code, Antigravity, Continue, Windsurf, JetBrains AI, and OpenCode environments. Discovers configurations, audits memory tiers, detects skill extraction state, analyzes behavioral tool chains, checks policy governance (v0.40+), and suggests reusable skills with evidence gating.…
scan
Scan your AI coding tool ecosystem — Gemini CLI, Claude Code, Antigravity (Desktop, CLI, IDE), Continue, Windsurf, JetBrains AI, OpenCode. Produces a maturity score, advisory recommendations, and optionally generates reusable SKILL.md files from your conversation patterns. Use when the user asks to audit their…
image-prompt
막연한 요청을 gpt-image-2(Codex $imagegen) 완성 프롬프트로 컴파일하는 스킬. 검증된 규칙 — 티어드 네거티브(기본 전부 긍정형+극소수 화이트리스트 2종), 앞 브래킷 금지·끝 AR 토큰만, 장비는 결과로 환원, HEX 명시, 1행=1컷, 사이즈락 6종, C1C12 플레이북, 화보 Format B(플랫 콤마형), 시네마틱 키아트(C11·그림자서사 shadownarrative), 프레젠테이션/슬라이드 덱(C12), 룩 프리셋 9종(홍대 인디 L9 포함), 홍보판촉물 그래픽 문법 P1P8(타이포-마스크·타이포-환경·오클루전·컬러블로킹 캠페인·메타…
media-analysis
음성/이미지/영상 파일 분석 워크플로: ffprobe 메타데이터 → 자막 우선 → 프레임 추출(네이티브 비전 분석) → whisper.cpp 전사 → tesseract OCR. Triggers: media, stt, transcribe, ocr, video analysis, youtube, 음성, 전사, 영상 분석.