List apps. Returns: [{id, name, mode}]

Select app for subsequent operations

Create app and auto-initialize workflow. Returns: {id, name, mode, start_node_id}

Initialize workflow draft (auto-called by create_app)

View workflow structure. Returns: {nodes:[{id,type,title}], edges:[{source,target}]}

Check for errors. Returns: {valid, errors[], warnings[]}

Publish draft as new version

Add node. Returns: {node_id}. Use get_node_schema first for config format.

Update node (PATCH-style, only include changed fields)

Delete node and its edges

Get node details

List nodes. Returns: [{id, type, title}]

Add multiple nodes at once. More efficient than multiple add_node calls.

Add documentation note to canvas (visual only, not executed). Text supports newlines for paragraphs.

Update sticky note

List sticky notes

Delete sticky note

Create edge between nodes

Remove edge between nodes

List edges. Returns: [{id, source, target}]

REQUIRED before add_node. Lists available node types.

REQUIRED before configuring a node. Get config schema for node type.

List AI models for LLM nodes. Returns: [{provider, model, model_type}]

List external tool providers

List tools from a provider with full schemas for agent configuration. Returns tool names, parameters (with form type, required, defaults).

List agent strategy providers for agent nodes. Returns providers with strategy names. Use get_agent_strategy for parameters.

Get agent strategy details including required parameters (instruction, query, model, tools, etc.) for agent_parameters config.

List knowledge bases for RAG

Get dataset details

List documents in dataset

Get document details

Query knowledge base (hit testing). Returns: [{score, content, document_name}]

Get workflow features config

Update features (partial)

Get environment and conversation variables

Execute single node for testing