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 Emily2040/nano-banana-image-skill --skill model-aware-image-prompt-engineergit clone --depth 1 https://github.com/Emily2040/nano-banana-image-skillWrote 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/emily2040/nano-banana-image-skill/model-aware-image-prompt-engineer)<a href="https://agentmods.dev/skills/emily2040/nano-banana-image-skill/model-aware-image-prompt-engineer"><img src="https://agentmods.dev/badge/skills/emily2040/nano-banana-image-skill/model-aware-image-prompt-engineer/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/emily2040/nano-banana-image-skill/model-aware-image-prompt-engineer"><img src="https://agentmods.dev/badge/skills/emily2040/nano-banana-image-skill/model-aware-image-prompt-engineer.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00210 | $0.01834 |
| Opus 5 | $0.00105 | $0.00917 |
| Sonnet 5 | $0.00042 | $0.00367 |
| Haiku 4.5 | $0.00021 | $0.00183 |
Grade A, and why
model-aware-image-prompt-engineer 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 12d 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 — 230 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Model-Aware Image Prompt Engineer
This skill writes prompts for the actual target model and surface instead of using one generic prompt style.
The core rule:
Route first. Write second.
An image prompt is not portable by default. Gemini, Midjourney, OpenAI, FLUX, Qwen, Z-Image, Stable Diffusion, Pony, and local ComfyUI checkpoints use different prompt languages, controls, and failure patterns.
First Decision
Before writing any prompt, identify:
- Target model or model family.
- Target surface.
- Task type.
- Input image roles.
- Whether the surface supports negative prompts.
- Whether the surface supports parameters outside the prompt.
- Whether exact text must appear in the image.
- Whether safety false positives are likely.
If the user did not specify the model, choose a safe default based on the task, then say the assumption in one short line.
Reference Loading
Load only the references needed for the request.
references/model-router.md: use for target model, surface, and task routing.references/prompt-formats.md: use for prompt templates, JSON-style briefs, tag prompts, negative prompts, API parameters, and edit prompts.references/model-cards-commercial.md: use for OpenAI, Gemini, Midjourney, BFL, Stability, Runway, Ideogram, Firefly, Recraft, Luma, Seedream, and hosted platforms.references/model-cards-open-local.md: use for Z-Image, Qwen-Image, FLUX local, Stable Diffusion, SDXL, SD3.5, Pony, Illustrious, NoobAI, Animagine, HunyuanImage, HiDream, OmniGen2, Sana, PixArt, Kolors, and Chroma.references/safety-and-false-positives.md: use for rejected prompts, policy friction, benign rewrites, and moderation ambiguity.references/evaluation-and-iteration.md: use for testing, seed grids, prompt diffs, and output review.examples/prompt-packs.md: use when the user asks for examples or a full prompt pack.
Evidence Labels
Use evidence labels whenever a model trick is not official.
Allowed labels:
What ships with it
12 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- agents/openai.yaml 258 B
- assets/hero.png 1628 KB
- assets/infographic.png 290 KB
- examples/prompt-packs.md 4.5 KB
- PR_BODY.md 2.5 KB
- README.md 8.6 KB
- references/evaluation-and-iteration.md 4.5 KB
- references/model-cards-commercial.md 9.0 KB
- references/model-cards-open-local.md 13 KB
- references/model-router.md 5.5 KB
- references/prompt-formats.md 5.9 KB
- references/safety-and-false-positives.md 4.9 KB
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.
- 12d ago First seen · 230 lines · 210 tokens per session scan A d4d8d023ae79
model-aware-image-prompt-engineer is a skill published in the GitHub repository Emily2040/nano-banana-image-skill (30 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 210 tokens to every session and 1,834 once invoked, about $0.0011 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 skills, from other repositories
image-prompt
A skill that turns a vague request into a detailed prompt for generating images, including options for covers, posters, promotional materials, typography, and other visual formats. A prompt is an instruction given to an image-generation system.
awesome-gpt-image-2
A structured library of reusable prompts for GPT-Image2, an image-generation model. It breaks prompts into parts such as subject, lighting, materials, composition, and visual details, with more than 20 templates.
mj-prompt
A prompt writer for Midjourney and niji, image-generation tools, using an East Asian visual style framework.
composition-to-prompt
A design skill that turns composition rules, such as the golden ratio and rule of thirds, into prompts for AI image generation.
ima2
Use the ima2-gen CLI/server to generate, edit, inspect, and manage local AI image generation jobs.
minimax-h3
Use when writing or debugging prompts for MiniMax H3 (Hailuo 3) video-with-audio generation, running the open weights locally in ComfyUI, choosing a quant or an acceleration LoRA for the VRAM you have, wiring reference-to-video with images, video or audio, or when a generated clip produces gibberish speech, drifts off…