Docs
Search docs…⌘K
DashboardStart free →
Getting Started
What is WyberAi?Your first buildHow credits work
Web Apps
Generating a web appConnecting SupabasePublishingCustom domainsExporting & self-hosting
Mobile Apps
Generating a mobile appThe live previewExporting to ExpoApp Store submission
Integrations
Claude & MCP
Account & Billing
Plans & pricingCredits explainedAI models
AI Agents›Running an agent

Running an agent

Once your canvas is built and tools are connected, you can run the agent with a single click and watch it execute each step in real time.

What you'll need
  • A built agent canvas
  • All required tools connected via OAuth

Step-by-step

  1. 1
    Open your agent canvas

    From the dashboard sidebar, click Agents and select the agent you want to run. The canvas opens.

  2. 2
    Click Run

    Click the Run button in the top bar of the canvas. If your agent has an input node (e.g. a start prompt), you'll be asked to fill it in before execution begins.

    Run button on canvasScreenshot placeholder — to be added after testing
  3. 3
    Watch the execution

    Each node lights up as it executes. An execution log streams in the bottom panel, showing each step's inputs and outputs in real time. The overall progress bar at the top shows how many steps have completed.

    Agent executing — nodes activeScreenshot placeholder — to be added after testing
  4. 4
    Review the output

    When the agent finishes, the final output is shown in the result panel on the right. If an action node sent data to an external service (created a GitHub issue, sent a Slack message), you'll see a confirmation link there.

    Agent output / resultScreenshot placeholder — to be added after testing
Note: If a node fails, execution stops at that step and the error is shown on the node. Fix the issue (re-connect the tool, adjust the prompt) and re-run.
← Connecting toolsNext: Bring your own keys →