cuozg/imagegen
skills/imagegen/SKILL.md
Use this skill when the user asks to generate, create, or edit images — concept art, product shots, icons, logos, textures, illustrations, or batch variants. Covers generation from text prompts and editing existing images (background removal/replacement, inpainting, style transfer, object removal). MUST use when the user says "generate an image," "create an icon," "make me a picture," "I need a hero image," "swap/remove the background," "edit this photo," "make it look like [style]," "batch of icons," "placeholder art," or describes needing any visual asset casually. Also triggers on: game assets (sprites, textures, UI icons), website visuals (hero images, blog headers), product photo edits, logo generation, infographics, or any request to produce an image from a description. Runs the bundled CLI (scripts/image_gen.py) and requires GEMINI_API_KEY or GOOGLE_API_KEY.
$ install --global
skillsauthnpx skillsauth add cuozg/oh-my-skills imagegenInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: May 13, 2026, 2:45 AM157.3s17 files scanned
SKILL.md
- name:
- imagegen
- description:
- >
- icons," "placeholder art," or describes needing any visual asset casually. Also triggers on:
- game
Image Generation Skill
Generate or edit images using Google Gemini Imagen API. Defaults: imagen-4.0-generate-001 (generate), imagen-3.0-capability-001 (edit). Use bundled CLI scripts/image_gen.py.
Decision Tree
- User provides input image OR says "edit/retouch/inpaint/mask/change only X" → edit
- Multiple different prompts/assets needed → generate-batch
- Otherwise → generate
Workflow
- Decide intent (decision tree above)
- Collect: prompt(s), exact text (verbatim), constraints/avoid list, input image(s)/mask(s)
- Batch: write temp JSONL to
tmp/imagegen/(one job per line), run once, delete after - Augment prompt via
references/prompt-augmentation.md(don't invent new requirements) - Pick aspect ratio (see table below)
- Run CLI:
python <skill-dir>/scripts/image_gen.py <command> <args> - Validate: file exists, size >10KB, report output path
- Iterate with ONE targeted change if result doesn't satisfy constraints
Aspect Ratios
| Use case | Ratio |
|----------|-------|
| App icon, avatar, logo, texture | 1:1 |
| Website hero, banner, cinematic | 16:9 |
| Mobile wallpaper, story, portrait | 9:16 |
| Blog header, product card | 4:3 |
| Infographic, poster | 3:4 |
CLI
python "<skill-dir>/scripts/image_gen.py" <command> <args>
# Or via run_skill_script:
run_skill_script(skill="imagegen", script="scripts/image_gen.py", arguments=["generate", "--prompt", "...", "--dry-run"])
Error Recovery
- Safety filter / no images returned: rephrase — less specific about people/violence. Add
--person-generation ALLOW_ADULTif appropriate. - Rate limit (429): CLI has built-in retry. Wait 60s if still failing.
- Empty result: simplify prompt first to confirm API, then add complexity.
- Edit policy rejection: add explicit invariants ("change only background; keep person unchanged").
Always diagnose which part caused the issue — don't re-run same prompt.
Environment
GEMINI_API_KEYorGOOGLE_API_KEYfor generation- Edit operations: Vertex AI mode — set
GOOGLE_CLOUD_PROJECT+GOOGLE_CLOUD_LOCATION+ ADC - CLI loads
.envfrom:--env-file→<project>/.env.local→<project>/.env→~/.env.local→~/.env - Never ask user to paste the key — ask them to set it as env var or
.envfile
Rules
--forcewhen re-running to overwrite previous attempt- Use Google GenAI Python SDK (
google-genai) — never raw HTTP - Prefer bundled CLI over writing one-off scripts
- Temp files:
tmp/imagegen/· Final output:output/imagegen/
References
references/prompt-augmentation.md— taxonomy, spec template, augmentation rulesreferences/cli.md— commands, flags, recipesreferences/image-api.md— API params, models, edit reference imagesreferences/prompting.md— prompting principles, iteration patternsreferences/sample-prompts-*.md— copy/paste prompts by category (generate, edit, website, game, asset-types)
Related Skills
cuozg/unity-image-gen
tools
VerifiedTrustedCommunity
Generate Unity raster image assets through Unity MCP: game sprites, item art, backgrounds, UI icons, portraits, concept images, transparent cutouts, image edits, upscales, background removal, and Unity scene or Game View screenshots. Use when a Unity project needs image files imported under Assets or screenshots captured from the editor. Do not use for meshes, audio, animation, materials, gameplay code, UI Toolkit layout, or generic non-Unity image generation.
4SKILL.mdUpdated May 29, 2026
cuozg/unity-technical
tools
VerifiedTrustedCommunity
Create Unity technical solution documents from user requirements, feature ideas, bug goals, specs, or codebase problems. Use when the user asks for a technical approach, architecture, implementation strategy, solution options, feasibility analysis, system design, or "how should we build/fix this" for Unity runtime, Editor, tools, assets, data, UI, WebGL, SDKs, or production pipelines.
4SKILL.mdUpdated May 26, 2026
cuozg/unity-mcp-orchestrator
tools
VerifiedTrustedCommunity
Orchestrate Unity Editor via MCP (Model Context Protocol) tools and resources. Use when working with Unity projects through MCP for Unity - creating/modifying GameObjects, editing scripts, managing scenes, running tests, or any Unity Editor automation. Provides best practices, tool schemas, and workflow patterns for effective Unity-MCP integration.
4SKILL.mdUpdated May 21, 2026
cuozg/goal-todo
development
VerifiedTrustedCommunity
Convert a spec document into an implementation TODO list in the same spec folder. U se when the user says goal-todo, todo from spec, generate tasks from spec, turn this spec into todos, create implementation checklist, extract tasks, or asks to read a Docs/Specs design doc and produce what must be implemented. Includes UI/UX review and codebase investigation before writing the checklist. Do not use for implementing the tasks, creating new goal files, writing test cases, or verifying completed work.
4SKILL.mdUpdated May 21, 2026