← Files Visual SandboxARCHIVED FILE

plugin.json

8.79 KB · Nov 5, 2026 · 01:19 UTC

↓ Download file

{
  "$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
  "name": "visual-sandbox",
  "version": "1.0.8",
  "description": "Generate images, video, audio, 3D and vector with Visual Sandbox. Adds the hosted MCP server and the prompt-craft skills for every model family.",
  "author": {
    "name": "Visual Sandbox",
    "email": "hello@visualsandbox.com",
    "url": "https://visualsandbox.com"
  },
  "homepage": "https://visualsandbox.com",
  "repository": "https://github.com/visualsandbox/skills",
  "license": "SEE LICENSE IN LICENSE",
  "keywords": [
    "ai image generator",
    "ai video generator",
    "text to video",
    "image to video",
    "animate photo",
    "image editing",
    "product photography",
    "product photos",
    "ugc ads",
    "video ads",
    "ad creatives",
    "ad variations",
    "ab testing ads",
    "tiktok videos",
    "instagram reels",
    "youtube shorts",
    "voiceover",
    "text to speech",
    "music generation",
    "sound effects",
    "svg",
    "logo design",
    "icons",
    "landing page design",
    "nano banana",
    "seedance",
    "veo",
    "kling",
    "flux",
    "3d model",
    "video editor",
    "captions",
    "consistent character",
    "talking avatar",
    "voice cloning",
    "short film",
    "podcast clips"
  ],
  "extensions": {
    "com.openai": {
      "interface": {
        "displayName": "Visual Sandbox",
        "shortDescription": "Make videos, ads & voiceovers",
        "longDescription": "Make videos, ads, voiceovers and images in ChatGPT with more than 50 AI models in one place. Animate a product photo, turn a script into a short film with the same character in every shot, or make a creator-style video ad with a voice and captions.\n\nMake a photo talk with a stock voice or your own cloned voice. Cut a long video into short vertical clips with word-timed captions. Turn one product photo into a set of store photos, or make several versions of one ad to test against each other.\n\nEdit in the same chat: join clips, add a voiceover, music and captions, trim to 9:16 and place your logo. Save a person or a product as a character once, and keep the same face and voice in every shot. Make music, sound effects, SVG logos, 3D models and web pages too.\n\nA larger run waits until you press Generate in the card. Every result shows in the chat and lands on a canvas at visualsandbox.com, where you see the whole set side by side and share a link. For help, contact hello@visualsandbox.com.",
        "developerName": "Visual Sandbox",
        "category": "Creativity",
        "capabilities": [
          "Generate videos from text or images",
          "Animate photos",
          "Short films with a recurring character",
          "UGC-style video ads",
          "Talking avatars from a photo",
          "Voice cloning",
          "Long videos cut into short clips",
          "Join and edit video clips",
          "Word-timed captions",
          "Voiceovers and text to speech",
          "Generate images from text",
          "Edit images with a reference photo",
          "Product photo shoots",
          "Ad creatives and ad variations",
          "Reusable characters",
          "Music and sound effects",
          "SVG logos and icons",
          "3D models",
          "Web pages and cards in editable layers",
          "Asks before a larger run"
        ],
        "websiteURL": "https://visualsandbox.com/mcp-server/",
        "supportURL": "https://visualsandbox.com/support/",
        "privacyPolicyURL": "https://visualsandbox.com/privacy/",
        "termsOfServiceURL": "https://visualsandbox.com/terms/",
        "defaultPrompt": [
          "Animate this product photo into a 5-second vertical video.",
          "Make a 10-second vertical video ad for my product with a script and a voiceover.",
          "Turn this product photo into a clean studio shot and two lifestyle photos."
        ],
        "brandColor": "#C99A00",
        "brandColorDark": "#FFD23D",
        "composerIcon": "./assets/icon.png",
        "logo": "./assets/icon.png"
      },
      "onboardingSkill": "./skills/vsb-setup/SKILL.md",
      "review": {
        "test_cases": {
          "positive": [
            {
              "description": "Text to image. Needs a demo account with credit.",
              "prompt": "Generate an image of a cozy coffee shop interior in warm morning light.",
              "tools_triggered": "an image_* model tool such as image_gpt_image_2_5_flare, show_media (get_skill when the model reads a prompting pack first)",
              "expected_behavior": "One image_* model tool call, then the Visual Sandbox card shows the finished image of the described scene, with a link to open it on visualsandbox.com. No extra paid variations."
            },
            {
              "description": "Read the account balance. Nothing is generated or charged.",
              "prompt": "How much credit is left on my Visual Sandbox account?",
              "tools_triggered": "get_balance",
              "expected_behavior": "States the balance in US dollars. Nothing is generated, nothing is charged, and no purchase link is offered."
            },
            {
              "description": "Product photo from an attached reference image. Needs image credit.",
              "prompt": "Turn this product photo into a clean studio hero shot on a white background.",
              "tools_triggered": "upload_media, an image_* model tool such as image_nano_banana_2, show_media",
              "expected_behavior": "Uploads the attached photo, runs one image edit with it as the reference, and the card shows one hero shot in which the product keeps its shape, colour and label.",
              "file_attachment_urls": [
                "https://cdn.visualsandbox.com/predictions/a8a2164f-3cd9-4666-a1c9-f0099b3e9d3a/76907e18-486a-4341-b9e9-6a03f4039a30/224f935f-c99b-4f44-bf66-f1820dc53646.jpg"
              ]
            },
            {
              "description": "Video with the price shown first. Needs video credit.",
              "prompt": "Make a 5-second vertical video of a paper airplane flying through a bright studio. Tell me the price first.",
              "tools_triggered": "estimate_cost, a video_* model tool such as video_p_video_2, show_media",
              "expected_behavior": "Says the price first. The card shows the request with its price in dollars and a Generate button; nothing runs or is charged before the reviewer presses Generate. After the press, the card shows the percent while the clip renders, then plays the finished 9:16 clip. A running job is not reported as a finished video."
            },
            {
              "description": "Stock voice narration. Needs speech credit.",
              "prompt": "Read this line in a warm, friendly voice: Welcome to Visual Sandbox, your studio for images, video and sound.",
              "tools_triggered": "list_voices, audio_elevenlabs_tts, show_media",
              "expected_behavior": "Picks a stock voice, keeps the words exactly, and the card plays the finished audio."
            }
          ],
          "negative": [
            {
              "description": "Written copy only. The request names Instagram, but asks for no media.",
              "prompt": "Write three short Instagram captions for a coffee shop that launches an oat milk latte."
            },
            {
              "description": "A general question about video specs. The person wants an answer, not a generated file.",
              "prompt": "What is the best aspect ratio and length for a YouTube Short?"
            },
            {
              "description": "Finding an existing stock photo. Visual Sandbox makes new media and does not search stock libraries.",
              "prompt": "Find me a stock photo of a mountain sunset that I can use in a presentation."
            }
          ]
        },
        "commerce": false,
        "commerce_description": "Visual Sandbox sells nothing in ChatGPT. The ChatGPT server address has no checkout or top-up tool and returns no purchase link. Generations are charged to the person's existing Visual Sandbox account. When the balance is too low, the plugin says so and says that credit is added on visualsandbox.com.",
        "demo_recording_url": "https://cdn.visualsandbox.com/uploads/a8a2164f3cd94666a1c9f0099b3e9d3a/8036ef315c38449aa77106607b382342.mp4"
      },
      "publication": {
        "release_notes": "A guided first run: your photos, product, logo or website become saved anchors, then one preview frame, a first 5-second video and three voice takes. New skills: motion graphics from HTML, brand themes read from your website, swap a person into a video, and feedback. Reels now cut several parts of a long video or a video link in one run. Films keep their script and shots between chats. A larger run waits for your press of Generate in the card."
      },
      "apps": "./.app.json"
    }
  }
}

SHA-256: b6867eb10b451ed8289b69f5eddf59812b178e0fbe3cbf696b7f4682b2bef97c