Executions

When you hit Run, Nick validates your workflow, executes each node in the correct order, streams results to you in real time, and stores a complete record for later review.

Execution Lifecycle

Every execution follows this path:

1. Trigger — You click Run (or a schedule fires)
2. Validate — Nick checks the workflow before running anything:
   • Exactly one Start node exists
   • All nodes are reachable (no orphans)
   • Required fields are configured (exchange selected, LLM model chosen, etc.)
   • Credentials exist and are accessible
3. Execute — Each node starts as soon as every node feeding it has finished. Nodes that don't depend on each other run in parallel.
4. Result — The execution completes as Completed, Failed, or Cancelled.

How Nodes Execute

Nick follows the connections in your workflow: a node starts as soon as every node it depends on has finished. Watch how execution flows through the graph:

In this example:

• The trigger fires first
• BTC Price and ETH Price run in parallel (both depend only on the trigger)
• Analyze runs once both price nodes complete (it needs both outputs)
• Alert runs after analysis is done

This parallel execution makes workflows significantly faster when nodes don't depend on each other.

Node States

Each node moves through states during execution:

State	Visual	Meaning
Pending	Grey	Waiting to execute
Running	Blue glow + spinner	Currently executing
Completed	Green glow + checkmark	Finished successfully
Failed	Red glow + X	Encountered an error
Skipped	No highlight	Downstream of a false conditional branch

Real-Time Streaming

As your workflow runs, Nick streams updates to the editor via Server-Sent Events (SSE). You see:

• Node highlights update live on the canvas — blue while running, green on success, red on failure
• Execution logs appear in the bottom panel as each step completes
• Progress counter tracks how many nodes have finished (e.g., "3/5 nodes")

Data Flow

Data passes between nodes through their connections. When a node completes, its output becomes available to downstream nodes as input.

For example, the Price Data node outputs price information. The next node can access it via the connection label — like price_data.prices[0].current in a template field.

Click any completed node in the execution view to inspect its inputs (what it received) and outputs (what it produced).

Failsafe Mode

Any node can be marked as failsafe. When a failsafe node fails:

• The error is logged but execution continues
• Downstream nodes still run (they receive null for that input)
• The node shows a dimmed red indicator instead of stopping the workflow

This is useful for non-critical steps like notifications — you don't want a Slack webhook failure to kill your entire trading strategy.

Retries

When a node fails, Nick tries it again before failing the run, unless running it again could repeat what the node did or would only fail the same way. Each retry shows in the execution logs: the failed attempt ends with "Trying again.", and the next one starts with "Attempt 2".

• Nodes that act (exchange and prediction-market orders, swaps, Aave and Lido, notifications, Storage) and Function nodes, whose code can act on its own, run once and are never retried, so a retry can't place a second order or send a message twice. If the code sandbox can't be reached at all, Nick connects again, since your code hasn't run. That holds even when one only reads, such as Storage set to Retrieve, or a Kalshi or Polymarket order node reading your balance or positions.
• Every other node (Price Data, LLM and the other data nodes, Conditional, Loop, Aggregator) is tried up to 5 times, waiting 10, 20, 40 and then 80 seconds between attempts. A rate limit is retried too.
• Errors you need to fix aren't retried, since they would fail the same way again: invalid settings, missing or rejected credentials, not enough credits, a request the service rejected, an error in your own Function code, or a node that took longer than 5 minutes.

A node that reads and takes longer than 5 minutes is stopped. One that acts (an order, a message) may still go through after that, so it fails asking you to check whether it did before running it again. A run that takes longer than 30 minutes is stopped and marked Failed. If a Nick server restarts during a run, the run carries on where it left off: nodes that already finished don't run again. A node that was in the middle of placing an order or sending a message when its server went down isn't run again either, since it may already have taken effect: it fails with "This step stopped unexpectedly and we couldn't confirm whether it finished," so you can check your account before running it again.

Cancellation

You can stop a running workflow at any time by clicking the Stop button. The run shows Stopping… while it winds down:

• No new nodes start. Function, Custom Function, data, notifications, storage and other non-trading nodes stop waiting for their result. This normally takes a few seconds. It does not undo a request already sent, and code already running in the sandbox may continue until it finishes or reaches its time limit.
• Only trading and DeFi nodes finish first so their result is recorded: Exchange, Polymarket Order, Kalshi Order, Aave, Cow Swap and Lido. This applies to the whole node type, whatever action it is configured to perform.
• The run is then marked as Cancelled in the history. If only the run's final trading steps were still running when you pressed Stop, the run ends as they do: Completed, or Failed if one of them fails.

Auto-Iteration

When the Workflow Assistant is open with auto-iterate enabled, Nick can automatically fix and retry failed executions:

1. Workflow runs and fails
2. The assistant analyzes the error
3. It modifies the workflow (adjusts config, fixes connections)
4. Reruns the workflow automatically
5. Repeats up to 5 times or until success

Execution History

Every execution is stored with a complete audit trail. Navigate to a workflow's Execution History to see:

• List view — all past runs with status, duration, and timestamp
• Flow tab — visual graph replay showing which nodes ran and their status
• Logs tab — full structured logs, searchable and downloadable
• Results tab — JSON output from each node
• Credits tab — cost breakdown (LLM tokens, API calls)

Next Steps

• Scheduling — set up automatic execution on intervals or cron schedules
• Workflows — learn how to build and wire workflows
• Nodes — explore all available node types