nano-banana-prompts
Generate optimized prompts for Gemini 2.5 Flash Image (Nano Banana). Use for image generation, crafting photo prompts, art styles, or multi-turn editing workflows with best practices.
77
66%
Does it follow best practices?
Impact
97%
1.03xAverage score across 3 eval scenarios
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./plugins/nano-banana-prompts/skills/nano-banana-prompts/SKILL.mdQuality
Discovery
89%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a solid skill description that clearly identifies its niche (Gemini 2.5 Flash Image prompt generation) and provides explicit trigger guidance. The trigger terms are natural and the description is distinctive. The main weakness is that the specific capabilities could be more granular—listing concrete actions like 'structure multi-turn editing chains' or 'apply style modifiers' rather than broad categories.
Suggestions
Add more concrete specific actions beyond categories, e.g., 'structure negative prompts, set aspect ratios, chain multi-turn edits, apply style modifiers' to improve specificity.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Names the domain (prompt generation for Gemini 2.5 Flash Image) and mentions some actions like 'crafting photo prompts', 'art styles', 'multi-turn editing workflows', but these are more categories than concrete specific actions. It doesn't list granular capabilities like 'apply negative prompts', 'set aspect ratios', or 'optimize token usage'. | 2 / 3 |
Completeness | Clearly answers both 'what' (generate optimized prompts for Gemini 2.5 Flash Image) and 'when' ('Use for image generation, crafting photo prompts, art styles, or multi-turn editing workflows'). The 'Use for...' clause serves as an explicit trigger guidance. | 3 / 3 |
Trigger Term Quality | Includes strong natural trigger terms: 'image generation', 'photo prompts', 'art styles', 'Gemini 2.5 Flash', 'Nano Banana', 'multi-turn editing'. These cover terms users would naturally say when seeking help with Gemini image generation prompts. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive due to the specific model name 'Gemini 2.5 Flash Image (Nano Banana)' and the narrow focus on prompt optimization for that particular model. Unlikely to conflict with general image editing or other AI prompt skills. | 3 / 3 |
Total | 11 / 12 Passed |
Implementation
42%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill provides genuinely useful, actionable prompt templates and examples for Gemini image generation, which is its core strength. However, it is severely bloated with repetitive content - the same advice appears in multiple sections (core principle, common pitfalls, common issues, prompt validation, troubleshooting table all repeat the same points). The entire document could be cut to roughly 1/3 its size without losing any information, and the remaining content should be split across referenced files for progressive disclosure.
Suggestions
Deduplicate heavily: the 'keyword vs sentence' advice, text rendering limits, multi-turn editing guidance, and style consistency warnings each appear 3-4 times. Consolidate each into a single authoritative location.
Split into multiple files: keep SKILL.md as a concise overview with core template and quick checklist, then reference separate files like PHOTOGRAPHY.md, ART_STYLES.md, TROUBLESHOOTING.md, and EXAMPLES.md.
Remove content Claude already knows: explanations of what aspect ratios mean (1:1 = square), what DOF is, basic photography concepts like three-point lighting. Just list the values/options without definitions.
Cut the 'Prompt Validation', 'Testing Your Prompts', and 'Common Issues & Solutions' sections which largely restate the same information already covered in 'Common Pitfalls' and the troubleshooting table.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | At ~350+ lines, this skill is extremely verbose and repetitive. The same advice (use complete sentences, not keywords; keep text under 25 chars; one edit at a time) is repeated 3-4 times across different sections. The prompt validation checklist, troubleshooting guide, testing section, and common issues sections all heavily overlap. Claude doesn't need to be told what aspect ratios are or basic photography concepts like DOF. | 1 / 3 |
Actionability | The skill provides concrete, copy-paste-ready prompt templates with filled-in examples for every category (photography, art, product, landscape, abstract, text). The templates use clear placeholder syntax and each is accompanied by a complete worked example. | 3 / 3 |
Workflow Clarity | The multi-turn editing workflow is clearly sequenced with good examples, and the refinement workflow has explicit steps. However, there are no real validation checkpoints with feedback loops - the 'prompt validation' is just a checklist of things to include, not a verify-fix-retry cycle. The testing section describes evaluation but lacks concrete validation commands or automated checks. | 2 / 3 |
Progressive Disclosure | This is a monolithic wall of text with no references to external files. The content could easily be split - templates in one file, troubleshooting in another, examples in another. Instead, everything is inline in a single massive document, making it hard to navigate and consuming excessive context window. | 1 / 3 |
Total | 7 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Repository
- secondsky/claude-skills
- Commit
88da5ff
- Reviewed
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.