NoriaFlow is built to catch mistakes early. Before a workflow ever runs in production.
The live linter
As you build, a built-in linter checks your workflow continuously. It flags problems like unknown variables, disconnected nodes, and missing required fields: before runtime, right in the editor. Fix the warnings and you avoid most failures.
Test a single node
You don't have to run the whole workflow to check one step:
- Double-click a step to open Studio, or select it and open its inspector, then choose Test.
- Review its captured, pinned or example input. Choose Edit simulation input to change the sample; Restore captured or example input restores the original.
- Click Run simulation and inspect the step result. Writes are simulated: no message is posted and no role or moderation action is performed. Services may still be read.
This makes iterating on a message template or an API call fast and safe.
Test the workflow, then run it for real
The editor topbar's Test action simulates an eligible trigger branch of the current draft. Choose the trigger and sample event, inspect each step's output and fix any errors. This is a workflow simulation, separate from Run simulation for a single step. Neither publishes your changes or proves that a bot has the permissions to post to a real channel.
To use real actions, choose Publish, review the preflight and confirm Publish live. Then use a supported manual Run or produce the actual event the workflow listens for. A real run can send messages, call APIs and change roles, and it counts toward your monthly runs. Use a dedicated test destination for the first real run.
Trace an execution
Open Activity → Runs and select an execution to inspect its status, inputs, outputs, errors and step timeline. Follow Review failed step to return to a failing step in the editor.
Open Activity → Analytics for aggregate metrics:
- Execution count, success rate, and average runtime.
- Metrics for the selected workflow and period. The execution details remain in Runs.
Analytics is computed live from the same execution history. It isn't a separate dataset with its own retention.
How long logs are kept
Every execution is stored for 90 days, then removed automatically. During the open beta, that 90-day window applies to every account. After the beta, the free plan's history will be trimmed to the most recent 7 days; paid plans keep the full 90.
When things fail
Failed nodes retry automatically with exponential backoff. Runs that exhaust their retries are kept so nothing disappears silently. You can inspect exactly what happened and why.
Need a hand reading a trace? Share it in our Discord and the community will help.
Setup checks and next steps
- Check the draft before activating it; saving edits and publishing them are separate actions.
- Resolve missing parameters and connection errors in the workflow that reports them.
See the extension setup guide for adding and reusing services, the node reference for parameters, and testing and debugging for execution diagnostics.