Give your AI agenta UGC video studio
UnrealUGC plugs into Claude Cowork, Claude Code, ChatGPT, OpenAI Codex, OpenClaw, Hermes Agent, and any other MCP-compatible AI agent. One sentence in, ready-to-post UGC ad out.
Any agent.Any host. Any model.
If it speaks MCP, UnrealUGC works with it. No proprietary SDK, no special-cased adapters.
Universal MCP install
{
"mcpServers": {
"unrealugc": {
"command": "npx",
"args": ["-y", "@unrealugc/mcp"],
"env": { "UNREALUGC_API_KEY": "usk_live_..." }
}
}
}Same JSON works in Claude Cowork, Claude Code, OpenClaw, Hermes Agent — anywhere MCP is supported. For remote clients (ChatGPT via MCP, Codex Cloud), point at https://unrealugc.com/api/mcp.
Hand the keys to your agent.Watch ads roll in.
UGC video creative is the slowest piece of most ad pipelines. With UnrealUGC and an agent driving it, that bottleneck disappears.
Generate ads on a schedule
Wire UnrealUGC into a cron-driven agent. Every morning, kick off three new ad variations for whichever product needs new creative.
Batch across creators
Ask the agent to produce one ad with each of your top 10 creators. It rate-limits itself, polls for completion, and reports back when all are ready.
Chat-driven creative
Describe an ad. The agent picks the model, voice, and aspect ratio, estimates the cost, then generates. Refine iteratively in the same conversation.
Plug into multi-step pipelines
Combine UnrealUGC with research agents, copywriting agents, and ad-platform agents. UGC video is the final step in a fully autonomous pipeline.
A video servicethat thinks like an agent.
Most video APIs were built for humans clicking buttons. We rebuilt ours for agents that need to reason about cost, schedule jobs, and recover from failure.
Built for agents from day one
Every endpoint is a typed MCP tool with JSON Schema input. Agents auto-discover capabilities — no fragile string parsing.
Async-first
Generate kicks off in <100ms and returns a job id. Use HMAC-signed webhooks so the agent gets a push the moment the video is done.
Cost-aware
estimate_video_cost tells the agent exactly how many credits a job will cost before spending them. No surprises on your bill.
Common questionsabout agent video generation
- Which AI agents can generate videos with UnrealUGC?
- Any MCP-compatible agent: Claude Cowork, Claude Code, Claude.ai remote MCP, OpenAI Codex, OpenClaw, Hermes Agent, and custom agents built on any framework. Agents without MCP support can use the REST API directly — same keys, same capabilities.
- Is the setup different for each agent?
- It's the same MCP server everywhere — only the config format differs per client (JSON for Claude Desktop, TOML for Hermes, a tools directory for OpenClaw). Each takes about two minutes; we keep per-agent guides for the popular ones.
- How do I keep an agent from overspending credits?
- Three layers: the estimate_video_cost tool tells the agent the exact credit cost before it generates, your credit balance is a hard ceiling, and you can give the agent its own API key and revoke it anytime.
- How does the agent know when a video is done?
- generate_video returns a job id immediately. The agent either polls get_video until status is COMPLETED, or you register an HMAC-signed webhook and the agent gets a push with the final video URL.
- Do agent-generated videos cost more?
- No. API, MCP, and webhook access are included on every plan, and agent-generated videos draw from the same credit balance as the web app.