VOOPA Render Check
Your video is a signal. Render Check is a Signal Labs tool that an agent can call after a generated video misses its brief and before anyone pays for another take. It returns timestamped evidence and recommends whether to edit the take, try one specific change, switch approach, or review the evidence again.
This developer preview runs locally. It uses the calling host’s visual review or its reviewer’s observations. The local preview can read an existing fal video result. The hosted World Engine review is in development.
Connect to the hosted preview
Use https://voopa-render-check.onrender.com/mcp as a Streamable HTTP MCP endpoint. The preview is free and rate limited. The hosting service sleeps after inactivity, so the first request may take about a minute. The hosted service accepts a completed fal result URL and returns frames to the calling agent, which supplies observations to the retry tool. The calling agent interprets the frames. The hosted endpoint does not run an independent vision model.
Use Render Check with fal
Connect the server and add the pre-retry instruction to your agent workflow. The guide includes the free endpoint, the optional paid endpoint, and the credit checkout.
Run the preview
Requirements are Node.js 22 or later and FFmpeg with ffprobe. Run:
The test generates its own one-second clip. The server listens at http://127.0.0.1:50951/mcp and accepts only local connections. For a process-based MCP host, launch node /absolute/path/to/mcp.mjs instead.
Pass a direct HTTPS video URL from an existing fal result as falVideoUrl, or pass the filename of an authorized local clip in the intake folder as file. A fal result is fetched from the supported fal media hosts, sampled, and the temporary clip is removed. This uses no generation credits.
Connect a workflow
The test verifies the MCP protocol with an independent client. The hosted provider integrations above are documented routes and still need live verification.
Tell the agent when to use it
When a paid video take misses the brief, use VOOPA Render Check before proposing another generation. Reuse the existing clip, prompt, and attempt history. Show the timestamp and one actionable recommendation in chat. Hold spending if the evidence is unclear.
Start with review_completed_render. Give it the clip or the video.url from the fal result, the must-haves, and prior attempts. A compatible process-based host can use model sampling. Over the local HTTP preview the host receives frames and calls decide_render_retry with its observations. Text-only hosts can use trusted observations supplied by a reviewer or separate vision component.
What this preview proves
The same tool interface works over local process and Streamable HTTP connections. The included smoke test exercises an inspect-and-decide flow without credentials or generation costs. The policy and evidence can be tested independently of a specific model. Nothing in the tool starts a render or authorizes spending.
