Actions overview

Actions are the operations that make things happen in Cargo. They’re the individual steps that enrich data, write to CRMs, send messages, apply logic, and connect to external services.

Action categories

Cargo organizes actions into two groups: workflow actions for building logic in tools and plays, and connectors for integrating with external systems and AI.

Workflow actions

These actions are available in the Tool and Play editors for building automation logic:

Logic

Control workflow behavior with branching, looping, filtering, and delays.

Code

Run custom JavaScript or Python code for advanced transformations.

AI

Use LLMs to generate content, extract information, and search documents semantically.

Sales

Purpose-built actions for lead allocation, scoring, and revenue operations.

Storage

Read, write, and search data in your Cargo data models.

Universal actions

These actions are available everywhere in Cargo—in tools, plays, agents, and via API:

Integrations

Connect to CRMs, enrichment providers, communication tools, and external platforms.

Tools

Call other Cargo tools from within your workflows to compose complex automations.

Agents

Invoke AI agents to handle complex, multi-step reasoning tasks autonomously.

Where actions are used

Context	Description
Tools	Build reusable workflows by connecting actions visually in the tool editor
Plays	Automate data-driven workflows that trigger when data model records change
Agents	Give AI the ability to select and execute actions autonomously
MCP	Expose your Cargo tools to any MCP-compatible AI system (Claude, Cursor, etc.)
API	Call tools and actions programmatically from your own applications

Learn more about leveraging actions in different contexts in Using Actions.

Retry policy

Some actions can retry automatically when they hit transient failures (rate limits, timeouts, temporary outages). In tools and plays, open the node options menu and select Retry policy. This setting is available for connector actions that use your own credentials; credit-based connectors follow the integration defaults. Retry settings:

Maximum attempts: Total attempts including the first run (set to 1 to disable retries).
Initial interval: Wait time before the second attempt, in seconds.
Backoff coefficient: Multiplies the delay after each attempt. Delay = initial interval * backoff coefficient^(attempt - 1).
Defaults: Leave a field empty to use the integration defaults.

Logic

Logic actions control how your workflow executes—branching based on conditions, looping through arrays, or applying filters.

Logic actions are available in Tools and Plays only.

Action	Purpose	Use when…
Balance	Distribute records across multiple routes proportionally	You need even workload distribution
Branch	Route execution based on a Yes/No condition	You have a single condition to evaluate
Delay	Pause execution for a specified time	You need to wait for external processes or rate limits
Filter	Stop execution if a condition is false	You want to gate workflow continuation
Group	Execute a sub-workflow for each item in an array	You’re processing multiple records in parallel
Human Review	Pause for human approval via Slack	You need manual approval before proceeding
Split	Randomly route to one of two branches	You’re A/B testing or sampling
Switch	Route based on multiple conditions	You have several possible paths
Variables	Create and store values for later use	You need to maintain state across nodes

Code

Code actions let you run custom scripts when built-in actions don’t cover your specific transformation, calculation, or logic requirements.

Code actions are available in Tools and Plays only.

Action	Purpose	Use when…
JavaScript	Run custom JavaScript	You need custom logic or transformations
Python	Run custom Python code	You prefer Python for data transformations

Code actions execute arbitrary code. Use caution with external inputs and always validate data before processing.

AI

AI actions leverage large language models to process, generate, and understand content.

AI actions are available in Tools and Plays only.

Action	Description
File search	Search through files and documents using semantic understanding. Find relevant content based on meaning, not just keywords.
Model ask	Query an LLM with a prompt. Generate text, analyze data, extract information, or make decisions.

AI actions consume credits based on token usage. Check the node picker for current pricing per 1K tokens.

Sales

Sales actions are purpose-built for revenue operations, replicating complex routing and scoring logic.

Sales actions are available in Tools and Plays only.

Action	Description
Allocate	Distribute leads to team members based on rules. Supports round-robin, territory-based, and capacity-weighted distribution.
Scoring	Create scoring models to evaluate records. Determine lead priority and potential value based on business rules.

Learn more about lead distribution in the revenue organization guide.

Storage

Storage actions interact with data models in your Cargo workspace, enabling data persistence and retrieval.

Storage actions are available in Tools and Plays only.

Action	Description
Memory	Key/value store for temporary data during a workflow run
Model custom column	Set custom column values on existing records
Model record	Add new records to your data models
Model search	Query your data models with filters and conditions

Storage actions work with your Cargo data models. Make sure you’ve set up your models before using these actions.

Memory

The Memory action provides a key/value store for persisting data across nodes. Use it to save values early in your workflow and retrieve them later—without passing data through every intermediate node. Scope options:

Scope	Behavior
Workspace	Data is shared across all plays in your workspace
Play	Data is isolated to the current play

Available actions:

Action	Description
Get	Retrieve a value by its key
Set	Store a value at a key (overwrites existing)
Get or Set	Return existing value, or set a default if none exists
Increment	Add to a numeric value (useful for counters)
Decrement	Subtract from a numeric value
Remove	Delete a key and its value

Configuration options:

Property	Description
Scope	Choose between `Workspace` (shared) or `Play` (isolated).
Key	A unique identifier for the stored value. Use descriptive names like `original_company_name` or `enrichment_result`. Supports expressions.
Value	(Set/Get or Set only) The value to store. Can be any data type—strings, numbers, objects, or arrays. Supports expressions.

Common use cases:

Preserve original data — Save input values before transformations so you can reference them later
Build counters — Use Increment to track how many times a play has run or count leads processed
Cache enrichment results — Store API responses to avoid duplicate calls within the same workflow
Share state across plays — Use Workspace scope to share data between different plays
Implement rate limiting — Track request counts with Increment/Decrement

Use the Increment action to build simple counters—track how many times a play has run, count leads processed today, or implement rate limiting logic.

Best practice: Use clear, descriptive key names that indicate what’s being stored and why. This makes your workflows easier to understand and debug.

Model Search

Search and retrieve records from any data model or dataset in your workspace. This action queries your data and returns matching records that you can use in subsequent steps. Common use cases:

Use case	Example
Find related records	Look up all contacts at a company when a deal is created
Check for duplicates	Search for existing leads before creating new ones
Aggregate data	Retrieve all recent activities for a given account
Cross-reference	Match records between different data sources

Configuration:

Property	Description
Model	The data model to search (e.g., Contacts, Companies, Deals)
Filters	Conditions that records must match
Limit	Maximum number of records to return
Sort	Order results by a specific field

Use expressions in your filter values to create dynamic searches. For example, filter by {{nodes.start.company_domain}} to find records related to the triggering record.

Model Record

Create new records in compatible data models directly from your workflows. This action allows you to write structured data back to models that support record creation. Supported data models:

Model type	Support
HTTP models	✅ Full support for creating records
All others	❌ Not supported

Data models synchronized from external services like CRMs don’t support creating records with this action. Use the dedicated write actions for each service (e.g., HubSpot Create Contact, Salesforce Create Lead) to ensure proper data sync.

Data type behavior: When you create a new record, the first value written to each field determines its data type. This type cannot be changed afterward, so plan your schema carefully before writing production data.

Model Custom Column

Extend synchronized data models with additional columns without modifying the source system. Custom columns let you store Cargo-specific metadata, flags, or computed values alongside your CRM or warehouse data. Examples of custom columns:

Lead scoring values calculated by your workflows
Enrichment timestamps for tracking data freshness
Processing flags like needs_review or outreach_sent
Aggregated metrics from other data sources

Configuration:

Property	Description
Model	The data model to extend with a custom column
Record ID	The primary identifier of the record to update
Column name	Name for the custom column (creates it if it doesn’t exist)
Value	The data to write to this column

Finding the correct Record ID:

Scenario	How to get the ID
Same data model as play trigger	Use `{{nodes.start._id}}`
Different data model	Use a Model Search action to find the record, then reference its ID
Known external ID	Use the external system’s ID (e.g., HubSpot contact ID)

If you’re updating records in the same data model that triggered your play, {{nodes.start._id}} is the simplest approach. For cross-data-model updates, always use a Model Search action first to locate the target record.

Storage best practices

Use Model Search to validate before writing

Before creating or updating records, search for existing entries to prevent duplicates. This is especially important when processing data from multiple sources.

Store enrichment metadata in custom columns

Track when records were last enriched, which providers were used, and whether the enrichment succeeded. This helps you avoid re-processing records unnecessarily.

Scope memory appropriately

Use play-scoped memory for play-specific state (like deduplication within a run) and workspace-scoped memory for cross-play coordination (like global rate limiting).

Handle empty search results

Model Search actions can return zero results. Add a Branch action to check for empty results and handle them gracefully—either skip processing or create a new record.

Integrations

Integration actions connect Cargo to external platforms. Each integration provides a set of actions specific to that platform.

Common actions

Action	Description
Insert	Create new records in the connected system
Find	Retrieve a specific record by ID
Match	Check if a record exists based on criteria
Search	Find multiple records based on filters
Update	Modify existing records
Upsert	Insert or update based on matching criteria
Delete	Remove records from the connected system
Associate	Link two records together (e.g., contact to company)

Tools

Call other Cargo tools from within your workflows. This lets you compose complex automations from smaller, reusable building blocks. Use cases:

Modular design — Build small, focused tools and combine them into larger workflows
Reuse logic — Share common operations (like enrichment or CRM writes) across multiple plays
Abstraction — Hide complexity behind a simple tool interface

When a tool is used inside another tool or play, it runs as a sub-workflow with its own inputs and outputs.

Agents

Invoke AI agents from within your workflows. Agents can handle complex, multi-step tasks that require reasoning and decision-making. Use cases:

Research tasks — Have an agent gather and synthesize information
Dynamic decisions — Let the agent choose the best action based on context
Complex workflows — Delegate multi-step processes to an autonomous agent

Learn more about agents

Understand how to build and configure AI agents.

Browsing actions

Cargo offers 120+ actions across all categories. To explore them:

Open the Tool or Play editor
Click the + button to open the node picker
Browse by category or search for specific capabilities
Click any action to see its configuration options and documentation

Next steps

Using actions

Learn how to leverage actions across tools, plays, agents, and APIs.

Build a tool

Create your first workflow using the visual tool editor.

Build a play

Automate data-driven workflows triggered by data model changes.

Build an agent

Give AI the power to select and execute actions autonomously.

Get Started

Reference

Integration

Action categories

Workflow actions

Logic

Code

AI

Sales

Storage

Universal actions

Integrations

Tools

Agents

Where actions are used

Retry policy

Logic

Code

AI

Sales

Storage

Memory

Model Search

Model Record

Model Custom Column

Storage best practices

Integrations

Common actions

Tools

Agents

Learn more about agents

Browsing actions

Next steps

Using actions

Build a tool

Build a play

Build an agent

Get Started

Reference

Integration

​Action categories

​Workflow actions

Logic

Code

AI

Sales

Storage

​Universal actions

Integrations

Tools

Agents

​Where actions are used

​Retry policy

​Logic

​Code

​AI

​Sales

​Storage

​Memory

​Model Search

​Model Record

​Model Custom Column

​Storage best practices

​Integrations

​Common actions

​Tools

​Agents

Learn more about agents

​Browsing actions

​Next steps

Using actions

Build a tool

Build a play

Build an agent

Action categories

Workflow actions

Universal actions

Where actions are used

Retry policy

Logic

Code

AI

Sales

Storage

Memory

Model Search

Model Record

Model Custom Column

Storage best practices

Integrations

Common actions

Tools

Agents

Browsing actions

Next steps