Yollomi AI Image Generator (All Models)

AI image generator skill (image, image generation). Multi-model image generator for Yollomi to generate AI images via one unified API endpoint. Requires YOLLOMI_API_KEY.

Security Scan

The skill's code, runtime instructions, and requested credential (YOLLOMI_API_KEY) are coherent with an image-generation integration; only minor documentation/env inconsistencies were found.

Purpose & Capability

Name and description match the implementation: the code and SKILL.md call a unified Yollomi API endpoint to generate images and list models. The only required credential (YOLLOMI_API_KEY) is appropriate for this purpose and no unrelated credentials or binaries are requested.

Instruction Scope

SKILL.md stays within the stated scope (POST /api/v1/generate, GET /api/v1/models). Minor inconsistency: the documentation includes video examples and video modelIds, but index.ts explicitly throws an error if params.type === 'video' (video generation is disabled). This mismatch is likely an implementation / doc divergence rather than malicious scope creep.

Install Mechanism

No install spec is provided (instruction-only + source files included). Nothing is downloaded from arbitrary URLs or installed system-wide; risk from install mechanism is low.

Credentials

The skill requires only YOLLOMI_API_KEY which is proportional to the task. One minor issue: SKILL.md examples and scripts reference an optional YOLLOMI_BASE_URL, but requires.env in the metadata does not declare it; index.ts does not actually read YOLLOMI_BASE_URL (it hardcodes https://yollomi.com). The inconsistency could confuse users or lead to the script using a different base URL than the code — not a credential-exfiltration indicator, but worth clarifying.

Persistence & Privilege

The skill does not request always:true, does not modify other skills, and only exposes an invocation tool. Allowing autonomous invocation is the platform default and not, by itself, a red flag here.

Assessment

This skill appears to do what it says: call Yollomi's unified API to generate images and list models. Before installing: 1) Confirm the skill source/trustworthiness (no homepage is provided and the registry owner is unknown). 2) Provide only an API key scoped for image generation and rotate it if you later remove the skill. 3) Note the small doc/implementation mismatches: video is documented but disabled in code, and YOLLOMI_BASE_URL is used in examples/scripts but not declared in metadata nor used by index.ts — verify which base URL your agent will actually call. 4) If you want extra caution, run the included test script (scripts/test-api.sh) in an isolated environment to confirm behavior and the endpoint contacted. If you rely on strict audit trails, treat the API key as sensitive and avoid sharing it with untrusted skills.

Yollomi AI API Skill

Generates images and videos via the Yollomi API. All models use a single unified endpoint with different modelId parameters.

Setup

1. API Key: Set YOLLOMI_API_KEY (environment variable).

Notes:

• Video generation is temporarily disabled in this skill build.

Unified Endpoint

POST /api/v1/generate

Headers: Authorization: Bearer ${YOLLOMI_API_KEY} or X-API-Key: ${YOLLOMI_API_KEY}
Content-Type: application/json

Body:

• type (required): "image" or "video"
• modelId (required): Model identifier
• Additional params depend on model (prompt, imageUrl, etc.)

Response (image): { images: string[], remainingCredits: number }
Response (video): { video: string, remainingCredits: number }

List Models

GET /api/v1/models

Returns all available image and video modelIds.

Common Examples

Generate image (Flux):

curl -X POST "${YOLLOMI_BASE_URL:-https://yollomi.com}/api/v1/generate" \
  -H "Authorization: Bearer $YOLLOMI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"type":"image","modelId":"flux","prompt":"A cat in a hat","aspectRatio":"1:1"}'

Remove background:

curl -X POST "${YOLLOMI_BASE_URL:-https://yollomi.com}/api/v1/generate" \
  -H "Authorization: Bearer $YOLLOMI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"type":"image","modelId":"remove-bg","imageUrl":"https://example.com/photo.jpg"}'

Generate video:

curl -X POST "${YOLLOMI_BASE_URL:-https://yollomi.com}/api/v1/generate" \
  -H "Authorization: Bearer $YOLLOMI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"type":"video","modelId":"kling-2-1","prompt":"A cat walking in the rain"}'

Aspect Ratio (aspectRatio)

Supported aspect ratios for text-to-image models:

ratio	description
1:1	Square (default)
16:9	Landscape
9:16	Portrait

Image ModelIds

modelId	Credits	Required	aspectRatio
flux	4/img	prompt	1:1, 16:9, 9:16
flux-schnell	2/img	prompt	same as above
flux-2-pro	15/img	prompt	same as above
remove-bg	0	imageUrl	-
nano-banana	4	prompt	1:1, 16:9, 9:16
nano-banana-pro	15	prompt	same as above
flux-kontext-pro	4	prompt	same as above
z-image-turbo	1	prompt	width, height
imagen-4-ultra	6	prompt	same as above
image-4-fast	3	prompt	same as above
ideogram-v3-turbo	3	prompt	same as above
stable-diffusion-3-5-large	7/img	prompt	same as above
seedream-4-5	4	prompt	same as above
object-remover	3	image, mask	-
face-swap	3	swapImage, inputImage	-
image-upscaler	1	imageUrl, scale	-
photo-restoration	4	imageUrl	-
qwen-image-edit	3	image, prompt	-
qwen-image-edit-plus	3	image, prompt	-
virtual-try-on	3	clothImage, personImage	-
ai-background-generator	5	imageUrl	prompt

Video ModelIds

modelId	Credits
openai-sora-2	~50+
google-veo-3	10
google-veo-3-fast	9
google-veo-3-1	10
google-veo-3-2	10
google-veo-3-1-fast	9
kling-2-1	9
kling-v2-6-motion-control	7/sec
minimax-hailuo-2-3	9
minimax-hailuo-2-3-fast	9
bytedance-seedance-1-pro-fast	8
runway-gen4-turbo	varies
pixverse-5	9
wan-2-5-i2v	9
wan-2-5-t2v	9
wan-2-6-i2v	29
wan-2-6-t2v	29

Workflow

1. Generate image → POST /api/v1/generate with type: "image", modelId, and model params
2. Generate video → POST /api/v1/generate with type: "video", modelId, prompt, optional inputs
3. List models → GET /api/v1/models
4. 401/402 → Check API key and credits

Reference

Full model list and params: models-reference.md or GET /api/v1/models