Nodes
Every node type by category: data, flow, compute, AI, and integration, and why each exists.
Nodes are the steps in a workflow. They fall into a few categories. You connect them on the canvas and data flows along the edges.
Two node groups have their own pages: the starting node is covered in Triggers, and the app action nodes in Integrations. This page documents every other node: data, flow, compute, and AI.
Data
Read from Trodo.
| Node | Why |
|---|---|
| Query | A typed, read-only query over events, runs, spans, or users, with filters, a time window, and a limit. |
| SQL | A read-only SELECT against a read replica, scoped to your team. Non-select statements are rejected. |
Flow
Control the path.
| Node | Why |
|---|---|
| Set fields | Build a new object from expressions. |
| Filter | Pass an item through only if a condition is true. |
| If | Branch down a true or false edge. |
| Loop | Fan out over an array; downstream nodes run per item. |
| Wait | Pause for a fixed delay. |
| Stop | End the workflow with success or failure. |
| Call workflow | Run another workflow as a sub-step. |
Compute
Run logic.
| Node | Why |
|---|---|
| Python | Run Python in a sandbox (numpy, pandas, scipy, requests available; configurable packages and timeout). |
| HTTP request | Call any API with a method, headers, and body. |
AI
Reason and act with a model.
| Node | Why |
|---|---|
| AI agent | An LLM that reasons and calls tools in a loop. Configured with a model and tools, see AI agents. |
| LLM call | A single prompt to a model, no tools. |
| MCP tool | Call one named tool on a remote MCP server. |
The AI agent has its own configuration (model, tools, MCP, and the loop) covered in detail on the AI agents page.
Integration
Act on an external app. These are the per-provider action nodes (Slack, GitHub, Linear, Jira, Salesforce), covered in Integrations.
Next
- AI agents for the agent node in depth
- Integrations for app actions
- Triggers