Image To Image

Use when someone asks for image-to-image, reference-image generation, transform image with prompt, edit image with AI, or CLI-based image-to-image workflows...

Security Scan

The skill's requirements, instructions, and install step are coherent with an image-to-image CLI integration and do not ask for unrelated access.

Purpose & Capability

Name/description (image-to-image for ricebowl.ai) matches the declared requirements: a CLI binary (ai-media) and an API base URL + API key. Requiring an ai-media CLI and AI_MEDIA_API_KEY/AI_MEDIA_BASE_URL is expected for a CLI client that talks to a media-generation API.

Instruction Scope

SKILL.md contains only CLI usage examples (ai-media config set-key, models list, image generate/get) and guidance about preparing reference images. It does not instruct reading unrelated files or accessing other environment variables. It references local docs paths (../ai-media-cli/...) that don't exist in this bundle, but that is a documentation link rather than an instruction to access sensitive system locations.

Install Mechanism

Install uses an npm/node package (ai-media-generator) which provides the ai-media binary. npm installs are traceable and typical for CLI tools, but installing arbitrary npm packages carries the usual supply-chain risk — review the package and its npm/GitHub repo before installing.

Credentials

Only two env vars are required: AI_MEDIA_BASE_URL and AI_MEDIA_API_KEY (primary). These are proportional to a CLI that must know which API endpoint to call and with what key. There are no unexplained tokens or unrelated credentials requested.

Persistence & Privilege

always is false and the skill does not request system-wide persistence or modification of other skills. The default ability for the agent to invoke the skill autonomously is normal for skills; it is not combined with other concerning privileges here.

Assessment

This is an instruction-only skill that expects the ai-media CLI and an API key/base URL for a media service. Before installing or enabling it: 1) verify the npm package (ai-media-generator) and the GitHub repo to ensure you trust the publisher; 2) set AI_MEDIA_BASE_URL only to the official service endpoint (to avoid pointing the client at an arbitrary/hostile server); 3) limit the API key's scope/permissions if the service allows it; 4) be aware that once enabled the agent can call the ai-media CLI using your API key (so the key will be usable by the agent for image generation). If you don't trust the npm package or the repo, do not install the package or provide your API key.

用这个 skill 处理这些请求：

• image to image
• image-to-image
• reference image generation
• transform image with prompt
• edit image with ai
• ai image to image generator

Default Route

ricebowl.ai
  -> recharge credits
  -> create API key
  -> set key
  -> upload or prepare reference image URL
  -> models list --json
  -> choose an image-to-image model
  -> image generate
  -> image get

Recommended Template

ai-media config set-key <KEY>
ai-media models list --json
ai-media image generate \
  --model <MODEL> \
  --prompt "<edit instruction>, <style>, <lighting>" \
  --aspect-ratio 1:1 \
  --param size=1024x1024 \
  --wait

先把参考图准备好，再让 prompt 只描述你要怎么改它。 如果模型要别的编辑字段，也直接用 --param KEY=VALUE。

Core Commands

ai-media config set-key <KEY>
ai-media config show
ai-media models list --json
ai-media image generate --model <MODEL> --prompt <PROMPT>
ai-media image get --task-id <TASK_ID>

When To Load References

• 参数全表、默认值、输出行为：读 ../ai-media-cli/references/cli-commands.md
• 平台充值、生成 key、平台 onboarding：读 ../ai-media-cli/references/platform-onboarding.md