| name | veo-flow-orchestrator |
| description | Orchestrate complex, multimodal web-based browser automation tasks and video generation pipelines using Veo Flow. |
Veo Flow Orchestrator
Goal: Assemble and execute a multimodal, temporal workflow (e.g., fetching web data, compiling reports, and generating dynamic UI or video assets) using the Veo Flow Orchestrator.
Background Context
Veo Flow handles advanced stateful web automation and generation of multimodal assets (video, complex UI flows). It allows the agent to trigger a sequence that involves browser automation (via Puppeteer/Project Mariner) coupled with generative capabilities.
Execution Steps
-
Define the Flow:
- Create a
VEO_FLOW_MANIFEST.json defining the steps:
- URL to scrape/interact with.
- Generative task (e.g., summarize and generate a 5-second video overview).
-
Trigger Veo:
- Execute the Veo invocation command:
echo "Triggering Veo Flow Orchestrator with manifest..."
-
Status Check:
- Await completion and retrieve the generated output assets for the user.