image-gen

Generate images via Nano Banana (Gemini 2.5/3.1 Flash Image) on OpenRouter. Use when the user asks to draw, illustrate, render or generate any kind of picture/diagram/scene.

Quality

83%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

77%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

An actionable, well-organized API recipe with executable code and useful error-recovery guidance. The main weaknesses are mild redundancy between the quick recipe and full request shape, time-sensitive pricing left inline, and no progressive disclosure into reference files despite the skill's length.

Suggestions

De-duplicate the Quick recipe and "Request shape (full)" sections, or mark one as the canonical reference, to tighten conciseness.

Move the version/price-sensitive "Cost (observed)" figures into a clearly marked section or a reference file so they can be updated without bloating the main body.

Consider extracting the full request/response schemas and troubleshooting into reference files (e.g. REFERENCE.md, TROUBLESHOOTING.md) and linking from a leaner SKILL.md overview.

Dimension	Reasoning	Score
Conciseness	Mostly efficient and free of filler, but the Quick recipe and "Request shape (full)" overlap, and the "Cost (observed)" block carries time-sensitive pricing ("at the time of writing") that the guidelines say should penalize conciseness unless isolated in a deprecated section.	2 / 3
Actionability	Provides complete, copy-paste-ready Python using only the stdlib for both text-to-image and image-input editing, with concrete request/response shapes and exact field paths (message.images vs content).	3 / 3
Workflow Clarity	For a single API-call task the action is unambiguous (build request -> call -> decode data URI -> write file), and the Troubleshooting section supplies concrete error-recovery hints (empty images, 403, refusal, corrupt PNG).	3 / 3
Progressive Disclosure	The body is well-sectioned but everything lives inline in one ~145-line file with no external references; cost details, full request/response schemas, and troubleshooting could be split out, so it is not "appropriately split" across files.	2 / 3
	Total	10 / 12 Passed

Description

90%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, concise description that answers both what the skill does and when to use it, with natural trigger terms and a clear niche. It is held back only on specificity, since it states a single action rather than enumerating the full range (generate, edit, vary) the skill supports.

Suggestions

Add the editing/variation capability to the action list, e.g. "Generate or edit images via Nano Banana ...", to lift specificity.

Consider narrowing the "diagram" trigger (e.g. "picture/scene") or scoping it to generative imagery to reduce overlap with diagram/plotting skills.

Dimension	Reasoning	Score
Specificity	Names one concrete action ("Generate images") plus the mechanism ("via Nano Banana ... on OpenRouter"), but does not enumerate multiple distinct actions such as editing or varying images, so it is not comprehensive.	2 / 3
Completeness	Explicitly states both what it does ("Generate images via Nano Banana ... on OpenRouter") and when to use it via an explicit "Use when the user asks to ..." clause.	3 / 3
Trigger Term Quality	Covers natural user phrasings well — "draw, illustrate, render or generate any kind of picture/diagram/scene" — which are terms users would actually say.	3 / 3
Distinctiveness Conflict Risk	Targets a clear niche (generative image output via the Nano Banana family on OpenRouter) with distinct triggers; the only mild overlap is the word "diagram", but the generative-image framing keeps it distinguishable.	3 / 3
	Total	11 / 12 Passed

Validation

87%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata' field is not a dictionary	Warning
metadata_field	'metadata' field is not a dictionary	Warning

	Total	14 / 16 Passed

Repository: EverMind-AI/Raven
Path: demos/skill_retrieval/skills/image-gen/SKILL.md
Commit: 4c2eb70

Reviewed: about 13 hours ago

Table of Contents

Discovery Implementation Validation

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.