AI & LLM Nodes
Text Generation
Multimodal & Vision
Knowledge Base
Agent Nodes
Core Agents
Specialized Agents
Logic & Control Flow
Conditionals
Iteration
Subflows
Data & File Operations
File Reading
File Generation
Data Transformation
List Operations
Integration Nodes
Google Workspace
Microsoft 365
Collaboration Tools
Developer Tools
Social Media
Communication
Business Tools
Web & APIs
Input/Output Nodes
Utility Nodes
Code Execution
Database
The SQL Query nodes run only reads (SELECT, SHOW, DESCRIBE, a SELECT on information_schema) until you turn on Allow Write Operations. With it off, any other query is skipped: the node returns empty results with a warning and says why in its Not Run Reason output. On PostgreSQL, MySQL and Oracle the query also runs in a read-only transaction, so a write hidden inside a SELECT fails the node. The toggle guards against accidental writes and is not a security control; a read-only database user is what guarantees nothing changes. Flows saved with an earlier version of the node keep their old behaviour.
Max rows caps the rows returned (1000 by default, 0 for no limit). When a query returns more, the node keeps the first ones, shows a warning and says so in Truncated Reason. Besides Results (JSON, CSV or a Markdown table), each query outputs Rows as objects with numbers kept as numbers and dates as ISO 8601 text, Row Count, and Affected Rows for an INSERT, UPDATE or DELETE (0 when the database does not report it).
Manage Database Schema builds the statement from a form in General mode: Create Table from column names and types (optionally NOT NULL, PRIMARY KEY or UNIQUE), Alter Table to add, drop or rename a column, or Drop Table. Names keep their case, so on PostgreSQL a capitalised name must be quoted in later queries. Advanced mode runs raw DDL (CREATE, ALTER, DROP, TRUNCATE, COMMENT) and refuses anything else. The node outputs the SQL that ran; a change the database rejects fails the node with the database’s reason.
Example: a nightly flow runs SQL Query with
SELECT * FROM orders WHERE created_at > now() - interval '1 day' and loops over Rows; a second SQL Query with Allow Write Operations on marks them exported, and its Affected Rows confirms how many changed.
Specialized/Industry Nodes
Node Categories
Nodes are organized into these categories in the UI:- AI Text: Text generation, NLP, document Q&A
- Agents: Autonomous AI agents with tools
- Logic: Conditionals, loops, control flow
- Data: File operations, transformations, parsing
- Integration: External service connectors
- Input/Output: Workflow inputs and outputs
- Utility: Helper functions, debugging, time operations
Finding the Right Node
Use the node palette search in the workflow editor to find nodes by:- Name (e.g., “gmail”, “pdf”, “generate”)
- Capability (e.g., “email”, “document”, “api”)
- Integration (e.g., “google”, “slack”, “notion”)
Node Versioning
Some nodes have multiple versions:- V2, V3: Enhanced versions with new features
- Legacy: Older versions (marked in UI)
- We recommend using the latest versions for new workflows
Next Steps
AI & LLM Nodes
Deep dive into AI-powered text and document nodes
Agent Nodes
Learn about autonomous AI agents
Integration Guide
Set up and use integration nodes
Building Workflows
Learn how to combine nodes into powerful workflows
This catalog gives an overview of Noxus’s nodes. For detailed parameter documentation and examples, refer to specific node pages or use the in-app documentation.