Understanding Nodes
Nodes are the building blocks of FluidGrids workflows. Each node represents a specific action, transformation, or integration point in your automation.
FluidGrids ships a growing catalog of nodes you can place as steps: built-in nodes for flow control, data handling, and actions, plus store nodes for external services and AI. In the builder they live in the Node Blocks panel (built-in and installed nodes) and the Node Store (everything available to the workspace, with Install / Update / Uninstall). A read-only Node Catalog page at /workflows/nodes lists every node registered in the workspace with its status.
Separately, trigger node types start a workflow — Trigger Manually (API), Webhook, Schedule, Form, Error, When Executed by Another Workflow, and the RSS Feed, HTTP Poll, Database Poll and Email (IMAP) polling triggers. See Workflows for what each one does.
Built-in Nodes
The built-in nodes are always available in every workspace, with no installation or credentials.
Flow Control Nodes
- IF — branch on a condition
- Switch — route to one of several branches by value
- Loop — iterate over a collection
- Merge — combine multiple branches back into one
- Wait — pause execution for a delay or until resumed
- No Operation — a pass-through placeholder, useful for labelling a branch
- Error — fail the run deliberately with a typed error
- Subworkflow — call another workflow as a step
- Compare Datasets — diff two collections and route the differences
Data & Transform Nodes
- Set — assign or reshape fields on items
- Filter — keep only items that match a condition
- Sort — order items by one or more keys
- Limit — cap the number of items passed downstream
- Split Out — expand an array field into separate items
- Remove Duplicates — drop duplicate items by one or more key fields
- Aggregate — roll many items up into one
- Transform — apply custom transformation logic
- Validate — check items against a schema
- Code — run a custom JavaScript or Python step over the items in a hardened sandbox
Action Nodes
- HTTP Request — REST and generic HTTP calls with authentication, retries, and error handling
- Log — write structured log entries during a run
Store Nodes
Store nodes cover the services teams automate most often. A store node has one or more operations (Gmail alone has dozens); you pick the operation when you add the node, and most operations need a connection to the service:
- Developer tools — GitHub
- CRM & Sales — Salesforce, HubSpot
- Finance & Payments — Stripe, QuickBooks
- Support — Zendesk
- Communication & Email — Gmail, SendGrid, Slack, Google Chat
- Google Workspace — Google Sheets, Google Drive, Google Docs, Google Calendar
- Data — PostgreSQL, Airtable
- Reviews & Social — Twitter/X, Trustpilot, G2, App Store Connect
- Project management — Linear
- Platform — BigConsole DataSink
- AI & parsing — AI Agent, Anthropic Chat Model, OpenAI Chat Model, Structured Output Parser, PDF Extract
- AI tools — Calculator Tool, Code Executor Tool, HTTP Request Tool
See Integrations for details on each connector and AI Capabilities for the AI nodes.
Building Your Own Node
Two paths exist:
- Ask FluidGrids AI. The assistant's Build a node mode scaffolds a complete node package from a description and hands you a zip. See AI Assistant.
- Build it with the node SDK. A node is a self-contained service: a
manifest.jsondescribing its operations, a GraphQL schema, resolvers and handlers.pip install fluidgrids-node-sdk, thenfluidgrids-node new <name>generates a working one, andvalidate,smokeandpackagecheck it and produce the zip to publish. See Custom Nodes.
Getting a node into a workspace goes through the Developer Portal › Applications submission and review flow; once approved it appears in the Node Store.