Execute Workflow Manually

Manual execution is the most basic way for you to run your workflow - development and debugging, one-time business requirements, and regular and unscheduled reports all use this entrance. This article explains clearly "where to click, what each item in the dialog box is, and what to see after running."

Three Manual Execution Entrances

1. Click "Execute" in the Editor

Most commonly used. Suitable for you who are changing a workflow and want to try it out immediately.

2. Click "▶ Execute" on the Workflow List Card

Suitable for a workflow that is already stable and you don’t want to open the editor.

3. Batch Execution

Check multiple items at the top of the workflow list and click "Batch Execute". In the pop-up window:

• Fill in a shared input JSON and apply it to all workflows in this batch
• Set the maximum number of concurrencies; 0 means no additional restrictions, and greater than 0 means the upper limit of simultaneous operations
• Choose whether to enable monitoring; each workflow will still create independent execution records

Execution Dialog Breakdown

No matter which entry is entered, after clicking "Execute", a dialog box with the same structure will pop up:

Parameter (Variables) Area

Automatically display overridable inputs according to workflow.variables:

• Prefill with default value
• Variables with UI metadata display labels, help text, numeric controls, or units
• file / file_or_text type can upload files; ordinary variables are overwritten in text mode
• Continue to use the default value in the workflow when the override value is not filled in

Preview and start execution

"Preview" only generates a static plan, listing Agents, steps, dependencies, condition markers, and missing variables. It does not create executions, nor does it call LLM, tools, or third-party APIs. "Start execution" will actually run. Please preview it for the first time and then execute it with test data.

Monitoring and execution ID

The dialog can enable monitoring and accept a custom execution ID; leave it empty to let the platform generate one. A custom ID is useful when artifacts must be reused or located by a stable identifier. Confirm that it does not conflict with an existing execution before use.

Debug Switch (Pro+)

After checking, this execution will enter debug mode - you can set breakpoints, single-step execution, view variables, and modify variables. See details Breakpoint Debugging.

Behavior After Startup

After clicking "Start Execution":

1. The front end immediately creates an execution record (status PENDING), and the page automatically jumps to the execution details.
2. Enter RUNNING when the concurrency quota is available; wait in PENDING when the current concurrency limit is exceeded.
3. The execution details page receives incremental updates through WebSocket; it is downgraded to polling when reconnection fails.
4. The step timeline, status cards, logs, and related tabs are updated as the execution progresses.

Real-time events and logs

Execution details will display events by categories such as Agent, LLM, Tool, External Message, Strategy, and User Interaction. Common content includes:

• Step status, time taken, output, number of retries and errors
• LLM calls and token statistics, Agent messages and tool calls
• code step logs, external messages, user interactions and artifacts
• Breakpoints and debugging context when debug is enabled

The specific event name will change with the steps and runtime; when troubleshooting, prioritize filtering by steps and categories, and then view the original event content.

Cancel Midway

The "Cancel" button in the upper right corner of the execution details page can actively stop a RUNNING execution:

1. The platform records the cancellation request and stops this execution task
2. execution enters CANCELLED, steps that have not yet been started will no longer run.
3. Requests that have been submitted to the model provider or external system are not guaranteed to be withdrawn; side effects that have occurred before cancellation will not be automatically rolled back

How to See the Results After Running

Execution Details Page

• Steps — Timeline, each step status and step details; final state execution can be re-executed from the specified step
• Conversation/External Message/DAG/Replay — View this execution process from different perspectives
• Inference/Performance/Debugging — Displayed when there is corresponding data or debugging session
• Artifact — Preview, download or generate sharing link based on deployment capabilities
• Execution Log — Filter and search event logs by level

Token & Cost

The top summary bar displays: total time taken, total tokens (in / out), total cost (USD). Click to expand to see the individual consumption of each LLM step.

Common Operations

Run Again from a Certain Step in the Middle

After the execution enters the final state, click "Reexecute from this step" next to a step. The previously retained step results continue to be reused, starting from the selected step and running backwards. Make sure the external side effects of this step can be repeated safely before proceeding.

Export Execution Report

“Export Execution Process” at the top exports JSON or YAML containing the execution status, step results, performance information, and the workflow definition fetched at export time. Download artifact files separately from the “Artifacts” tab.

When should manual execution be used and when should other triggers be used?

• Manual — Development and debugging, one-time requirements, "I want it to run now"
• Scheduled Runs — Fixed frequency automatic triggering
• Webhook — External system event driven

Next

• Execution Monitoring and Logging — More detailed event flow/log/cost statistics
• Scheduled Runs — Let workflow run automatically
• Breakpoint Debugging — Locate problems when workflow results do not meet expectations