Node catalog

Nodes are the building blocks of a workflow. Drag them from the searchable palette, wire their inputs/outputs with {{variable}} references, and configure each in the context-aware panel. Categories below mirror the platform's task families. For how a node executes, see Task / Node Types.

Triggers

How a workflow starts (see Event Bus / Triggers).

Node	What it does	Key config
Webhook	Start from an HTTP POST (JSON, form-data, file upload)	Path, payload type, secret key, max file size, MIME filters
Scheduled	Start on a cron schedule	Cron expression, timezone, pause/resume, skip-if-running
Event Source	Start from an event (e.g. Drive folder watch)	Source, event-type filter, custom criteria
Manual	Run on demand	Input JSON

AI Operations

Node	What it does	Key config
Classification	Categorize a document (type + category, confidence)	Categories, model, multi-classification
Extraction	Pull structured fields using a prompt	Prompt (library or inline), output JSON schema, model
Summarization	Generate an executive summary	Summary type (brief/detailed/bullet), max word count
LLM	Free-form text generation	Prompt, model (Vertex/OpenAI/Anthropic), output (JSON/text), Data Shield
Combined Classify + Extract	Classify then conditionally extract in one node	Categories + per-category extraction
Agent	Run a goal-driven agent (Agents)	`agent_id` or inline goal, tools, model, max_iterations, temperature, conversational, Data Shield

Document

Node	What it does	Key config
OCR	Extract text from scanned PDFs / images	Auto language detection
Document Generation	Create DOCX from a template with data merge	Template, variable substitution

Knowledge

Node	What it does	Key config
RAG Enable	Index a document for semantic search	Collection, force reprocess
RAG Query	Query the knowledge base in natural language	Top-K, similarity threshold, reranking, session id

Integration

See Integrations.

Node	What it does	Key config
Email Send	Send email with `{{variable}}` substitution	to/CC/BCC, subject, HTML body, sender
Gmail	Read / reply with thread preservation	message/thread id, label filter
Google Drive Fetch	Retrieve a file from Drive by id (or search)	File id / search query (user OAuth)
REST API	Call an external HTTP API	Method, URL, headers, body, timeout, JSONPath output mapping
Web Scraping	Fetch and extract web page content	URL, HTML→markdown, Open Graph metadata

Control Flow

Node	What it does	Key config
Decision	Conditional branching with true/false routing	Operators (`==`, `!=`, `>`, `<`, `>=`, `<=`, `in`, `contains`, `matches`), JSONPath refs, edge labels
Wait / Poll	Poll an endpoint until a condition is met	Endpoint, poll interval, timeout, success/failure match
For-Each	Iterate over an array, per-item execution	Array ref, accumulator, current-item variable

Data Transform

Node	What it does	Key config
Variables	Reference any earlier node's output via `{{variable_name}}`	JSONPath field extraction
Input Mapping	Resolve variables from context before a node runs	`{{var}}` references
Output Mapping	Extract fields from a result into named variables	JSONPath expressions

Built-in context variables: {{user_id}}, {{execution_id}}, {{DMS_URL}}, plus all webhook form fields. File uploads produce {field}_document_id, {field}_file_url, {field}_mime_type.

The authoritative, always-current task-type list is GET /api/workflows/task-types and the synced feature reference. This catalog is curated for readability; the reference is regenerated weekly and is the source of truth.