Your AI Agent.Full Media Studio.
Connect Claude, Cursor, or any MCP-compatible agent to faktry. Generate images, animate videos, produce audio — all through natural language. No tab-switching.
70+ Creative Tools, Callable from Any AI Agent
The Model Context Protocol (MCP) makes faktry natively callable by your AI agent. Ask Claude to generate a product image, animate it into a cinematic video, or save results to your library — the agent picks the right tool, queues the job, polls for completion, and returns the result with an inline preview. No copy-pasting URLs. No switching apps. Your prompt is the only interface.
Two Ways to Connect
OAuth for one-click setup in browser-based tools. API key for developer tools and local IDEs.
One-Click OAuth
Paste the MCP server URL into your app, approve access on faktry, and authentication is handled automatically. No API key to manage.
Claude.ai · Cursor · ChatGPT Desktop · Hermes Agent
Connect with OAuth →Config File + API Key
Add faktry to your tool's MCP config file using an API key. Works with all MCP-compatible developer tools.
Claude Desktop · Claude Code · Windsurf · VS Code · Continue · Zed
{ "mcpServers": { "faktry": { "type": "http", "url": "…/api/mcp" } } }Get Setup Instructions →
What Your Agent Can Do
Real workflows powered by faktry's MCP tools — describe the task, the agent handles the rest.
Product Image Studio
Type "Generate 4 product mockups for our new sneaker on a white background" — Claude selects GPT-Image-1.5 or Seedream, generates all variants in parallel, and returns inline previews directly in the chat. No uploads, no switching apps.
Models
GPT-Image-1.5 · Seedream v4.5 · Nano Banana 2
Output
Inline preview in chat
Format
PNG · WebP · transparent BG
Cinematic Product Video
Share a product image and say "Turn this into a 16:9 cinematic video with camera movement" — Claude calls Veo 3.1 or Kling 3.0, queues the async job, polls every few seconds, and returns the finished video URL the moment it's ready.
Models
Veo 3.1 · Kling 3.0 · Seedance 2.0 · Happy Horse
Output
Async polling — automatic
Format
16:9 · MP4
UGC Reference Campaign
Attach 3 influencer + product reference images and say "Create 9:16 UGC clips for TikTok" — Claude uploads all references, generates clips with Kling 3.0 or Veo 3.1 Fast, and returns a set of clips ready for review.
Models
Kling 3.0 (4 refs) · Veo 3.1 (3 refs) · Happy Horse (9 refs)
Output
Multiple clips in one run
Format
9:16 · MP4
Social Reel Factory
Drop in any image and say "Make a 6-second vertical reel and save it to my social folder" — Claude picks Seedance 2.0 or Happy Horse, queues the job, delivers the reel, and calls faktry_library_save automatically.
Models
Seedance 2.0 · Happy Horse · Veo 3.1 Fast
Output
6s reel + auto library save
Format
9:16 · MP4
Every Frontier Model, One Connection
One MCP server gives your agent access to models from OpenAI, Google, ByteDance, Black Forest Labs, Ideogram, and more — including every major frontier release.
Access 70+ AI models across image, video, audio, and PDF. Your agent automatically selects the best model for each task — or you can specify one directly.
Video Models
Image Models
5 Built-in Workflow Prompts
Invoke these by name in any MCP-compatible agent to run a complete multi-step production workflow.
product_cinematic_videoProduct Cinematic Video
Product image → 16:9 cinematic video with audio. Automatically selects the best model from Veo 3.1, Kling, Seedance 2.0, or Happy Horse based on your brief.
ugc_reference_videoUGC Reference Video
Multi-ref images → 9:16 UGC-style clip. Supports up to 9 reference images across Veo 3.1, Kling 3.0, Grok, and Happy Horse.
product_image_studioProduct Image Studio
Text description → high-quality product image. Chooses from GPT-Image-1.5, GPT-Image-2, Seedream v4.5, or Nano Banana 2.
social_reel_from_imageSocial Reel from Image
Image → 6–8s vertical 9:16 reel optimized for Instagram Reels, TikTok, and YouTube Shorts.
product_reveal_storyboardProduct Reveal Storyboard
Start + end frame → smooth product reveal video using first-last-frame models from Veo 3.1 and Seedance 2.0.
Invoke by name in any MCP-compatible agent
Connect your agent70+ Tools Across Every Media Type
Your agent calls any of these natively — no custom integration, no API wrangling.
Image
20+ tools · generate, edit, upscale, remove BG
Video
20+ tools · generate, trim, merge, subtitle
Audio
9 tools · TTS, music, clone, transcribe
9 tools · merge, split, compress, extract
Library
list & save across sessions
Jobs
status polling & cancel
All async tools follow the same pattern: submit → poll faktry_job_status → receive result_url. The agent handles this loop automatically.
Add faktry to Your AI Agent
One MCP server. 70+ tools. Every frontier model. Start free with 100 credits — no card required.
Included with every account
Frequently Asked Questions
Which AI agents and tools support the faktry MCP server?
faktry works with any MCP-compatible client. OAuth one-click setup is supported by Claude.ai, Claude Desktop, Cursor, ChatGPT Desktop (Business/Enterprise/Edu), and Hermes Agent. Config file setup works with Claude Code, Windsurf, VS Code Copilot, Continue.dev, Kilo Code, Codex CLI, Zed, and any other tool that accepts remote HTTP MCP servers.
Do I need a separate API key for MCP?
It depends on your connection method. OAuth (Claude.ai, Cursor, ChatGPT Desktop) handles authentication automatically — no API key needed. Config file setup (Claude Desktop, Claude Code, VS Code, etc.) requires a faktry API key, which you generate once in your dashboard and paste into the config once.
How does the async job pattern work?
AI generation tools (image-to-video, text-to-video, image generation) are async — they return a job_id immediately. The agent polls faktry_job_status every few seconds until status changes to 'completed', then receives the result_url. This loop runs automatically inside the agent's tool-calling logic — you just describe what you want.
Can the agent show generated images inline in the chat?
Yes. When an image job completes, faktry returns the result as an MCP image content block — Claude Desktop and Claude.ai render it inline in the conversation. You see a thumbnail of the generated image directly in the chat, not just a URL. Video results return a URL link since inline video playback isn't supported by MCP clients yet.
What AI models can the agent access through MCP?
The agent has access to all models on faktry — Veo 3.1 and Sora 2 for video, Flux 2 Pro/Max, GPT-Image-1.5, GPT-Image-2, Seedream v4.5, Nano Banana 2, and Ideogram v3 for images, plus ElevenLabs for voice and Suno for music. The 5 named workflow prompts automatically select the best model for each task type.
Can the agent save results to my library automatically?
Yes. Just say 'save it to my library' or 'save it to the products folder' in your prompt and the agent calls faktry_library_save with the result URL, filename, and folder. Library files are persistent across sessions and retrievable via faktry_library_list — so your agent can reference assets from previous sessions.
Explore More Features
The same tools available through MCP are also accessible from the web app
Image Generation
20+ AI models: Flux 2, GPT-Image, Ideogram, Seedream and more.
Video Creation
15 operations: AI generation, editing, trimming, merging, conversion.
Audio Processing
9 operations: Transcription, TTS, voice cloning, mixing, conversion.
Workflow Pipelines
Automate multi-step pipelines — batch generate, process, and export.
Content Library
Manage all your generated media. Accessible by MCP agents across sessions.