菜单
Unified image-generation skill. Default provider is nano-banana-pro (Gemini 3 Pro Image). Supports text-to-image; edits via nano-banana-pro.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use this skill when exploring a fuzzy idea, thinking through a concept from first principles, stress-testing assumptions, or crystallizing what you're actually trying to build before committing to a plan. Load when asked to brainstorm, when a problem statement is vague, or when the user is saying "I'm thinking about..." rather than "build this."
Use this skill when verifying that implementation work meets acceptance criteria, reviewing code for security vulnerabilities, assessing code quality and stability, or producing a formal review report. Load when conducting QA, security audits, or acceptance testing against a defined spec.
Use this skill when verifying a build, preparing a deployment, creating pull requests, running health checks, or managing a release. Load when the task is to get code from "merged" to "running in production" — or to confirm it is ready to be merged.
Use this skill when writing implementation plans, spec files, or step-by-step execution plans for engineers or sub-agents. Load when translating an architecture document and requirements into a concrete, ordered execution plan where each step must be self-contained and independently executable.
Use this skill when writing product requirements documents, defining acceptance criteria, or translating research and user needs into structured, testable specifications. Load when the task is to define what the system must do — not how — including functional requirements, non-functional requirements, user stories, acceptance criteria, and scope boundaries.
Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, artifacts, posters, or applications (examples include websites, landing pages, dashboards, React components, HTML/CSS layouts, or when styling/beautifying any web UI). Generates creative, polished code and UI design that avoids generic AI aesthetics.
| name | image-generation |
| description | Unified image-generation skill. Default provider is nano-banana-pro (Gemini 3 Pro Image). Supports text-to-image; edits via nano-banana-pro. |
Generate a single image per request via a provider script. Default provider is nano-banana-pro unless the user asks for OpenAI.
If the user does not provide a prompt, you must generate a random prompt about a cartoon octopus image and pass it to the script.
Providers:
nano-banana-pro → src/skills/image-generation/providers/nano-banana-pro.pyopenai → src/skills/image-generation/providers/openai-image-generation.pyRun scripts using an absolute path (do NOT cd to the skill directory first). Always run from the user's working directory so outputs land where they are working.
If the user does not specify an API key, assume the runtime already has secrets loaded (e.g., via framework secret storage or environment loading). Do NOT read .env directly.
If you need to run a command in a shell that already loads environment variables, you can use:
bash -lc "set -a; source .env; set +a; <command>"
Generate new image (nano-banana-pro):
uv run /absolute/path/to/src/skills/image-generation/providers/nano-banana-pro.py --prompt "your image description" --filename "output-name.png" [--resolution 1K|2K|4K] [--api-key KEY]
Edit existing image (nano-banana-pro):
uv run /absolute/path/to/src/skills/image-generation/providers/nano-banana-pro.py --prompt "editing instructions" --filename "output-name.png" --input-image "path/to/input.png" [--resolution 1K|2K|4K] [--api-key KEY]
Generate new image (openai):
uv run /absolute/path/to/src/skills/image-generation/providers/openai-image-generation.py --prompt "your image description" --filename "output-name.png" [--size 1024x1024] [--quality high|standard] [--api-key KEY]
Goal: fast iteration without burning time on 4K until the prompt is correct.
uv run /absolute/path/to/src/skills/image-generation/providers/nano-banana-pro.py --prompt "<draft prompt>" --filename "yyyy-mm-dd-hh-mm-ss-draft.png" --resolution 1K--input-image for every iteration until you’re happy.uv run /absolute/path/to/src/skills/image-generation/providers/nano-banana-pro.py --prompt "<final prompt>" --filename "yyyy-mm-dd-hh-mm-ss-final.png" --resolution 4KNano Banana Pro (resolution):
Map user requests to API parameters:
1K1K2K4KOpenAI (size/quality):
--size default 1024x1024--quality default highGEMINI_API_KEY or --api-keyOPENAI_API_KEY or --api-key.env directlybash -lc "set -a; source .env; set +a; <command>"Prompt templates (use only when needed):
Generate new image (nano-banana-pro):
uv run /absolute/path/to/src/skills/image-generation/providers/nano-banana-pro.py --prompt "A cartoon octopus chef flipping pancakes" --filename "2026-02-02-09-15-00-octopus-chef.png" --resolution 4K
Edit existing image (nano-banana-pro):
uv run /absolute/path/to/src/skills/image-generation/providers/nano-banana-pro.py --prompt "make the octopus wear a sailor hat" --filename "2026-02-02-09-20-30-octopus-sailor.png" --input-image "original-octopus.png" --resolution 2K
Generate new image (openai):
uv run /absolute/path/to/src/skills/image-generation/providers/openai-image-generation.py --prompt "Cartoon octopus reading a map in a tiny submarine" --filename "2026-02-02-09-25-10-octopus-sub.png" --size 1024x1024 --quality high