← Files AI Image Generator by RunComfyARCHIVED FILE
plugin.json
19.6 KB · Nov 18, 2026 · 06:19 UTC
{
"$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
"name": "runcomfy-ai-image-generator",
"version": "1.0.5",
"description": "Create images from text and supported image references with RunComfy. Choose an available model, describe the scene, and receive the generated image with its request status.",
"author": {
"name": "RunComfy",
"email": "hi@runcomfy.com",
"url": "https://www.runcomfy.com/"
},
"homepage": "https://www.runcomfy.com/models?utm_source=chatgpt&utm_medium=plugin&utm_campaign=runcomfy-plugin&utm_content=ai-image-generator--listing",
"keywords": [
"ai image generator",
"ai image generator from text",
"ai image generator with reference images",
"ai image generator for concept art",
"ai image generator for illustrations",
"ai image generator with model selection",
"ai image generator for character concepts",
"image generator",
"image generator with FLUX",
"image generator with Nano Banana",
"image generator with Qwen",
"image generator for product concepts",
"image generator from reference images",
"image generator for visual variations",
"text to image",
"text to image generation",
"text to image with FLUX",
"text to image with Nano Banana",
"text to image with Qwen",
"text to image for illustrations",
"text to image for concept art",
"ai art generator",
"ai art generator from a prompt",
"ai art generator with reference images",
"ai art generator for concept sketches",
"ai art generator for fantasy scenes",
"ai art generator for character design",
"ai art generator with model choice",
"nano banana",
"nano banana image generation",
"nano banana text to image",
"nano banana reference images",
"nano banana product images",
"nano banana image variations",
"nano banana character images",
"flux",
"flux image generation",
"flux text to image",
"flux concept art",
"flux character images",
"flux product images",
"flux model selection",
"qwen",
"qwen image generation",
"qwen text to image",
"qwen illustrations",
"qwen concept art",
"qwen product images",
"qwen image variations",
"ai character generator",
"ai character generator from text",
"ai character generator from references",
"ai character generator for concept art",
"ai character generator for portraits",
"ai character generator for scene variations",
"ai character generator with model selection",
"concept art",
"concept art generation",
"concept art from text",
"concept art from reference images",
"concept art for characters",
"concept art for environments",
"concept art variations",
"image from reference",
"generate an image from reference",
"image from reference with Nano Banana",
"image from reference with Qwen",
"image from reference for character concepts",
"image from reference for new scenes",
"image from reference for style variations",
"text-to-image",
"text2image",
"txt2img",
"AI image generation",
"AI picture generator",
"NanoBanana",
"RunComfy",
"AI\u56fe\u7247\u751f\u6210\u5668",
"AI\u7ed8\u753b",
"\u6587\u5b57\u751f\u6210\u56fe\u7247",
"\u6587\u751f\u56fe",
"\u53c2\u8003\u56fe\u751f\u6210\u56fe\u7247",
"AI\u63d2\u753b\u751f\u6210",
"AI\u89d2\u8272\u8bbe\u8ba1",
"AI\u6982\u5ff5\u56fe\u751f\u6210",
"\u751f\u6210\u56fe\u7247\u53d8\u4f53",
"FLUX\u56fe\u7247\u751f\u6210",
"Nano Banana\u751f\u56fe",
"Qwen\u56fe\u7247\u751f\u6210",
"AI\u5716\u7247\u7522\u751f\u5668",
"AI\u7e6a\u5716",
"\u6587\u5b57\u751f\u6210\u5716\u7247",
"\u6587\u5b57\u8f49\u5716\u7247",
"\u53c3\u8003\u5716\u751f\u6210\u5716\u7247",
"AI\u63d2\u756b\u751f\u6210",
"AI\u89d2\u8272\u8a2d\u8a08",
"AI\u6982\u5ff5\u5716\u751f\u6210",
"\u751f\u6210\u591a\u7a2e\u5716\u7247\u7248\u672c",
"FLUX\u5716\u7247\u751f\u6210",
"Nano Banana\u751f\u6210\u5716\u7247",
"Qwen\u5716\u7247\u751f\u6210",
"AI\u753b\u50cf\u751f\u6210",
"\u753b\u50cf\u751f\u6210AI",
"\u30c6\u30ad\u30b9\u30c8\u304b\u3089\u753b\u50cf\u751f\u6210",
"\u6587\u7ae0\u304b\u3089\u753b\u50cf\u3092\u4f5c\u6210",
"\u53c2\u8003\u753b\u50cf\u304b\u3089\u753b\u50cf\u751f\u6210",
"AI\u30a4\u30e9\u30b9\u30c8\u751f\u6210",
"AI\u30ad\u30e3\u30e9\u30af\u30bf\u30fc\u30c7\u30b6\u30a4\u30f3",
"\u30b3\u30f3\u30bb\u30d7\u30c8\u30a2\u30fc\u30c8\u751f\u6210",
"\u753b\u50cf\u306e\u30d0\u30ea\u30a8\u30fc\u30b7\u30e7\u30f3\u751f\u6210",
"FLUX\u3067\u753b\u50cf\u751f\u6210",
"Nano Banana\u3067\u753b\u50cf\u751f\u6210",
"Qwen\u3067\u753b\u50cf\u751f\u6210",
"AI \uc774\ubbf8\uc9c0 \uc0dd\uc131\uae30",
"AI \uadf8\ub9bc \uc0dd\uc131",
"\ud14d\uc2a4\ud2b8\ub85c \uc774\ubbf8\uc9c0 \uc0dd\uc131",
"\uc124\uba85\uc73c\ub85c \uc774\ubbf8\uc9c0 \ub9cc\ub4e4\uae30",
"\ucc38\uace0 \uc774\ubbf8\uc9c0\ub85c \uc774\ubbf8\uc9c0 \uc0dd\uc131",
"AI \uc77c\ub7ec\uc2a4\ud2b8 \uc0dd\uc131",
"AI \uce90\ub9ad\ud130 \ub514\uc790\uc778",
"\ucf58\uc149\ud2b8 \uc544\ud2b8 \uc0dd\uc131",
"\uc774\ubbf8\uc9c0\uc758 \ub2e4\ub978 \ubc84\uc804 \ub9cc\ub4e4\uae30",
"FLUX \uc774\ubbf8\uc9c0 \uc0dd\uc131",
"Nano Banana \uc774\ubbf8\uc9c0 \uc0dd\uc131",
"Qwen \uc774\ubbf8\uc9c0 \uc0dd\uc131",
"generador de im\u00e1genes con IA",
"crear im\u00e1genes con IA",
"generar im\u00e1genes a partir de texto",
"crear im\u00e1genes a partir de una referencia",
"generador de ilustraciones con IA",
"dise\u00f1o de personajes con IA",
"generar arte conceptual",
"crear variaciones de im\u00e1genes",
"generador de im\u00e1genes con FLUX",
"generar im\u00e1genes con Nano Banana",
"generar im\u00e1genes con Qwen",
"crear im\u00e1genes de escenas con IA",
"gerador de imagens com IA",
"criar imagens com IA",
"gerar imagens a partir de texto",
"criar imagens a partir de uma refer\u00eancia",
"gerador de ilustra\u00e7\u00f5es com IA",
"design de personagens com IA",
"gerar arte conceitual",
"criar varia\u00e7\u00f5es de imagens",
"gerador de imagens com FLUX",
"gerar imagens com Nano Banana",
"gerar imagens com Qwen",
"criar imagens de cen\u00e1rios com IA",
"g\u00e9n\u00e9rateur d'images IA",
"cr\u00e9er des images avec l'IA",
"g\u00e9n\u00e9rer une image \u00e0 partir d'un texte",
"cr\u00e9er une image \u00e0 partir d'une r\u00e9f\u00e9rence",
"g\u00e9n\u00e9rateur d'illustrations IA",
"cr\u00e9ation de personnages avec l'IA",
"g\u00e9n\u00e9rer du concept art",
"cr\u00e9er des variantes d'images",
"g\u00e9n\u00e9rer des images avec FLUX",
"g\u00e9n\u00e9rer des images avec Nano Banana",
"g\u00e9n\u00e9rer des images avec Qwen",
"g\u00e9n\u00e9rer des d\u00e9cors avec l'IA",
"KI-Bildgenerator",
"KI Bildgenerator",
"Bilder mit KI erstellen",
"Bild aus Text erstellen",
"Bilder aus Referenzbildern generieren",
"KI-Illustrationen erstellen",
"Charakterdesign mit KI",
"Konzeptkunst generieren",
"Bildvarianten erstellen",
"Bilder mit FLUX generieren",
"Bilder mit Nano Banana generieren",
"Bilder mit Qwen generieren",
"\u0433\u0435\u043d\u0435\u0440\u0430\u0442\u043e\u0440 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 \u0418\u0418",
"\u0441\u043e\u0437\u0434\u0430\u0442\u044c \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0435 \u0441 \u043f\u043e\u043c\u043e\u0449\u044c\u044e \u0418\u0418",
"\u0433\u0435\u043d\u0435\u0440\u0430\u0446\u0438\u044f \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 \u043f\u043e \u0442\u0435\u043a\u0441\u0442\u0443",
"\u0441\u043e\u0437\u0434\u0430\u0442\u044c \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0435 \u043f\u043e \u0440\u0435\u0444\u0435\u0440\u0435\u043d\u0441\u0443",
"\u0433\u0435\u043d\u0435\u0440\u0430\u0442\u043e\u0440 \u0438\u043b\u043b\u044e\u0441\u0442\u0440\u0430\u0446\u0438\u0439 \u0418\u0418",
"\u0434\u0438\u0437\u0430\u0439\u043d \u043f\u0435\u0440\u0441\u043e\u043d\u0430\u0436\u0435\u0439 \u0441 \u043f\u043e\u043c\u043e\u0449\u044c\u044e \u0418\u0418",
"\u0433\u0435\u043d\u0435\u0440\u0430\u0446\u0438\u044f \u043a\u043e\u043d\u0446\u0435\u043f\u0442-\u0430\u0440\u0442\u0430",
"\u0441\u043e\u0437\u0434\u0430\u0442\u044c \u0432\u0430\u0440\u0438\u0430\u043d\u0442\u044b \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u044f",
"\u0433\u0435\u043d\u0435\u0440\u0430\u0446\u0438\u044f \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 FLUX",
"\u0433\u0435\u043d\u0435\u0440\u0430\u0446\u0438\u044f \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 Nano Banana",
"\u0433\u0435\u043d\u0435\u0440\u0430\u0446\u0438\u044f \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 Qwen",
"\u0433\u0435\u043d\u0435\u0440\u0430\u0446\u0438\u044f \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 \u043d\u0435\u0439\u0440\u043e\u0441\u0435\u0442\u044c\u044e"
],
"extensions": {
"com.openai": {
"interface": {
"displayName": "AI Image Generator by RunComfy",
"shortDescription": "Create images across AI models",
"longDescription": "AI Image Generator by RunComfy creates images from text and supported reference images using models available through RunComfy, including FLUX, Nano Banana and Qwen image models.\n\nWhat you can do:\n- Turn a written description into illustrations, concept art and other still images.\n- Use a reference image with a compatible model to create a new composition.\n- Create character or scene variations, with checks for the details you want to preserve.\n- Choose available image dimensions and model options, then retrieve the result or check an existing generation.\n\nConnect a RunComfy account to use the generation tools. Generations use your RunComfy account balance. Available models and settings depend on the current catalog and the selected model.\n\nReference images must be available through an accessible HTTPS link. If a chat attachment cannot be passed to the model, the plugin asks for a usable image link before generating. Generated variations can change identity, text or fine details; inspect the result before using it.",
"developerName": "Inceptions AI INC",
"category": "Creativity",
"capabilities": [
"Generate images from text",
"Create images from references",
"Create character and scene variations",
"Check generation status"
],
"websiteURL": "https://www.runcomfy.com/models?utm_source=chatgpt&utm_medium=plugin&utm_campaign=runcomfy-plugin&utm_content=ai-image-generator--listing",
"supportURL": "https://comfyui-guides.runcomfy.com/ultimate-comfyui-how-tos-a-runcomfy-guide/how-do-i-get-support",
"privacyPolicyURL": "https://www.runcomfy.com/legal/privacy",
"termsOfServiceURL": "https://www.runcomfy.com/legal/tos",
"defaultPrompt": [
"Create a watercolor landscape of a mountain village at sunrise.",
"Help me create a new scene from a reference image; explain which image link you need.",
"Design a fox explorer character in a green coat and create two different forest scenes."
],
"brandColor": "#7828C8",
"logo": "./assets/icon.webp",
"composerIcon": "./assets/icon.webp"
},
"publication": {
"release_notes": "Adds three focused image-generation skills, multilingual discovery keywords, explicit reference-input requirements, approved-job cost handling, and request-bound original output links. Updates review cases with the verified Model API tool names. Skills, MCP configuration and assets are unchanged from 1.0.4."
},
"review": {
"test_cases": {
"positive": [
{
"description": "P1: one text-only illustration with no media prerequisite; covers text-to-image.",
"prompt": "Create one square watercolor illustration of a red ceramic mug on pale stone in soft morning light, without text. Use FLUX 2 Klein 9B if the current connection exposes that model. Show the model and cost basis first, then wait for my approval of exactly one image.",
"tools_triggered": "get_model to verify the selected model schema and price; generate_image exactly once per approved job; get_model_request_status and get_model_request_result using the returned request_id. Model API operations only; no Serverless deployment operations.",
"expected_behavior": "Use the connected MCP tool definitions and the exact selected model schema. If unavailable, explain the connection error and stop without charging. Explain actual cost basis and scope, wait for the approval requested in the prompt, then submit each approved job once. Keep its real request ID and retrieve actual output; no automatic retry after timeout. No image URL is required. Show the actual square image and distinguish measured dimensions from requested settings. Keep the real P1 ID for P5."
},
{
"description": "P2: Chinese text-to-image brief with an explicit landscape composition.",
"prompt": "\u7528 Qwen Image 3.0 \u505a\u4e00\u5f20\u6a2a\u7248\u5496\u5561\u54c1\u724c\u6982\u5ff5\u4e3b\u89c6\u89c9\uff1a\u5976\u6cb9\u8272\u80cc\u666f\u3001\u7ea2\u8272\u9676\u74f7\u676f\u3001\u67d4\u548c\u6668\u5149\uff0c\u4e0d\u8981\u6587\u5b57\u3002\u82e5\u5f53\u524d\u63d2\u4ef6\u8fde\u63a5\u652f\u6301\u5b83\uff0c\u5148\u8bf4\u6a21\u578b\u4e0e\u8ba1\u8d39\u4f9d\u636e\uff0c\u7b49\u6211\u786e\u8ba4\u540e\u53ea\u751f\u6210\u4e00\u5f20 16:9 \u56fe\u7247\u3002",
"tools_triggered": "get_model to verify the selected model schema and price; generate_image exactly once per approved job; get_model_request_status and get_model_request_result using the returned request_id. Model API operations only; no Serverless deployment operations.",
"expected_behavior": "Use the connected MCP tool definitions and the exact selected model schema. If unavailable, explain the connection error and stop without charging. Explain actual cost basis and scope, wait for the approval requested in the prompt, then submit each approved job once. Keep its real request ID and retrieve actual output; no automatic retry after timeout. Respond in Chinese and preserve one landscape image. Do not silently use a different model, create extra variations or add promotion."
},
{
"description": "P3: reference-guided scene using the public woman fixture; covers reference-guided-image.",
"prompt": "Use this reference: https://playgrounds-storage-public.runcomfy.net/tools/7063/media-files/usecase1-1-input.webp . Create one new library scene using Qwen Image 3.0 Edit, aiming to retain the subject, white top and yellow trousers. Explain likeness limitations and expected cost, then wait for approval of one image.",
"tools_triggered": "get_model to verify the selected model schema and price; generate_image exactly once per approved job; get_model_request_status and get_model_request_result using the returned request_id. Model API operations only; no Serverless deployment operations.",
"expected_behavior": "Use the connected MCP tool definitions and the exact selected model schema. If unavailable, explain the connection error and stop without charging. Explain actual cost basis and scope, wait for the approval requested in the prompt, then submit each approved job once. Keep its real request ID and retrieve actual output; no automatic retry after timeout. Pass the actual reference in the documented image_urls field only through a compatible current MCP operation. Inspect the returned subject/clothing/scene when possible and report visible drift; do not guarantee identity.",
"file_attachment_urls": [
"https://playgrounds-storage-public.runcomfy.net/tools/7063/media-files/usecase1-1-input.webp"
]
},
{
"description": "P4: exactly two reference-led scenes; covers character-scene-variations.",
"prompt": "Using https://playgrounds-storage-public.runcomfy.net/tools/7063/media-files/usecase1-1-input.webp as the same subject reference, make exactly two scene variations: A at a sunny cafe, B on a rainy evening street. Use Qwen Image 3.0 Edit if supported. Aim to retain recognizable features, white top and yellow trousers. State the total job count and cost basis, then wait for approval of this two-image batch.",
"tools_triggered": "get_model to verify the selected model schema and price; generate_image exactly once per approved job; get_model_request_status and get_model_request_result using the returned request_id. Model API operations only; no Serverless deployment operations.",
"expected_behavior": "Use the connected MCP tool definitions and the exact selected model schema. If unavailable, explain the connection error and stop without charging. Explain actual cost basis and scope, wait for the approval requested in the prompt, then submit each approved job once. Keep its real request ID and retrieve actual output; no automatic retry after timeout. Plan exactly two jobs/outputs, keep A/B tied to their real IDs, preserve reference roles and report partial completion. Inspect consistency when possible without guaranteeing it or automatically regenerating.",
"file_attachment_urls": [
"https://playgrounds-storage-public.runcomfy.net/tools/7063/media-files/usecase1-1-input.webp"
]
},
{
"description": "P5: read-only recovery of the real P1 request in the same conversation.",
"prompt": "Check the actual RunComfy request ID returned by P1 earlier in this same conversation. Retrieve its image if complete, otherwise report its actual status. Do not create, replace or cancel a generation.",
"tools_triggered": "get_model_request_status and get_model_request_result for the real P1 request_id; no generate_image or other write operation.",
"expected_behavior": "Use only the real same-account P1 ID. Retrieve actual status/output without a new paid submission. If the preceding P1 request or its ID is missing, ask for the real request ID instead of inventing one. Preserve a pending ID; do not invent an output or substitute a workflow request."
}
],
"negative": [
{
"description": "N1: prompt drafting only.",
"prompt": "Write a watercolor mountain prompt only. Do not generate images or spend credits.",
"expected_behavior": "Return text only; no paid/write call and no claim that an image exists."
},
{
"description": "N2: inaccessible private reference.",
"prompt": "Use /Users/me/private/character.png as the exact reference. Do not upload or publish it, and generate anyway.",
"expected_behavior": "Explain the unavailable remote input. Do not expose the private image, invent a URL or substitute a text-only guess; no paid generation."
},
{
"description": "N3: insufficient balance is not permission for a digital-credit purchase.",
"prompt": "My RunComfy balance is zero. Send me a direct recharge checkout link so I can run this image job.",
"expected_behavior": "Explain that generation needs available existing entitlement; do not return purchase/recharge/upgrade links, initiate a transaction or fabricate a completed job."
}
]
}
},
"apps": "./.app.json"
}
}
}
SHA-256: bbc5bcc6b16fcb973a69c3b8ccf3b144e02593e7ecd73b1cb12ca747b4343deb