Actions
Actions are MCP tools in Boltic MCP. They serve as the bridge between AI applications and your systems, wrapping Boltic primitives (Workflows, Serverless) and third-party APIs with consistent inputs, outputs, authentication, and observability. Actions make it easy to expose your business logic, data sources, and integrations as standardized tools that AI models can discover and use.
By converting your existing Boltic resources into actions, you enable AI applications to interact with your systems through a unified interface. Each action provides a well-defined contract with typed parameters, clear descriptions, and structured outputs that help AI models understand when and how to use them. This standardization eliminates the need for custom integrations and allows you to leverage your existing infrastructure investments while enabling powerful AI-driven automation.
Add Action
Creating an action is a straightforward process that transforms your existing Boltic resources into discoverable MCP tools. Follow these steps to add a new action to your MCP server:
- In the MCP panel, click Actions tab and then click Add Action. A fly-out menu appears on the right side of the window, providing access to all available action types.
-
Select an Action Type from the available options. Each action type corresponds to a different Boltic primitive or integration. Refer to Action Types to understand the details and use cases for each type. The action type determines what resource you'll be exposing and how it will be invoked.
-
Configure your action by adding a clear, descriptive name and detailed description. The description is crucial as it helps AI models understand the purpose and functionality of your action. Then, configure the action parameters that define the input schema. Refer to Action Parameters for detailed information about parameter configuration and best practices.
- Review your configuration to ensure all required fields are properly filled and parameters are correctly defined. Once satisfied, click Save to create the action. The action will immediately become available in your MCP server and can be discovered by connected AI applications.
Action Types
Boltic MCP includes managed integrations you can expose as MCP actions.
Here are the categories of integrations available:
1. Workflows
Workflow actions enable you to expose your existing Workflows as MCP tools, allowing AI applications to trigger complex, multi-step business processes. This action type is ideal for orchestrating operations that involve multiple steps, conditional logic, data transformations, and integrations with external services.
2. Integrations
Integration actions allow you to connect AI applications directly to popular SaaS services, third-party APIs, databases, and external systems. With access to hundreds of pre-built integrations, you can quickly expose any service as an MCP action without writing custom code. See Workflow integrations for a complete list of available integrations.
3. Serverless
Serverless actions enable you to convert any Serverless function endpoint into an MCP action, providing maximum flexibility for custom logic and API interactions. This action type gives you fine-grained control over how requests are structured and processed.
4. Tables
Table actions allow you to expose your Boltic Tables data as MCP actions, enabling AI applications to interact with your data through secure, schema-aware operations. These actions provide a safe and structured way to perform CRUD (Create, Read, Update, Delete) operations on your tables.
5. External MCP
External MCP actions enable you to connect to any existing MCP server and run it behind Boltic MCP, creating a unified gateway for multiple MCP servers. This powerful feature allows you to aggregate capabilities from different MCP servers into a single, manageable interface.
All the above integrations allow you to create powerful and feature-rich MCP servers with minimal effort, enabling you to expose diverse capabilities through a unified interface.
6. Gateway
Gateway actions enable you to expose Boltic Gateway routes as MCP actions, allowing AI applications to interact with your API gateway endpoints. This action type provides a bridge between MCP and your existing gateway infrastructure, enabling you to leverage gateway features like request routing, authentication, and payload transformation within your MCP server.
Action Parameters
Parameters define the input schema for an action, serving as the contract between AI applications and your systems. They specify what inputs an action expects, including parameter names, descriptions, types, validation rules, and whether they're required or optional. Well-defined parameters are crucial for AI models to understand how to use your actions correctly and effectively.
Why Parameters Matter:
- AI Model Understanding: Clear parameter definitions help AI models determine when and how to invoke your actions
- Validation: Built-in validation rules catch errors before execution, improving reliability
When configuring parameters, provide detailed descriptions that explain not just what the parameter is, but also how it should be used and what values are expected. This context helps AI models make better decisions about parameter values and improves the overall quality of interactions.
Example
In the below given example, we have defined parameters to match the workflow’s input contract: name, age, and email. These parameters tell the AI what values it must send when triggering the workflow, and they also let the system validate inputs before execution.
Mark name and email as required because the workflow needs them to identify the user and send the confirmation email to the provided address. Keep age required only if your workflow logic depends on it; otherwise, make it optional.
Write descriptions like instructions so the AI fills values correctly—for example: name as the user’s full name, age as a whole number in years, and email as a valid recipient email address (this is where the confirmation email will be sent).
Example
This action triggers the “Add to Cart” workflow. The Parameters section defines the exact inputs the AI must send to add a product to the user’s cart—so the workflow can reliably identify the item, set the quantity, and pick the right variant/size.
In this example, the action expects item_id, quantity, item_size, and anonymous_id. item_id is set as a required String because it’s the primary identifier of what to add. quantity should typically be a required number/integer so the workflow knows how many units to add. item_size is usually helps select the correct product variant—so if the size isn’t available or the item isn’t serviceable, the AI should ask the user to choose an alternative. anonymous_id is generally a String used to link the cart to an anonymous/guest session, and can be required or optional depending on how your cart session is handled.
JSON Schema
Here's an example of how to define parameters using JSON Schema:
[
{
"key": "item_id",
"value": "",
"type": "string",
"description": "The item ID to be added to the cart",
"required": true
},
{
"key": "quantity",
"value": "",
"type": "string",
"description": "The quantity to be added to the cart",
"required": true
},
{
"key": "item_size",
"value": "",
"type": "string",
"description": "The size of the item to be added to the cart. If no size can be be found or fetched then send OS",
"required": true
},
{
"key": "anonymous_id",
"value": "",
"type": "string",
"description": "The anonymous ID of the user generated for this session. Reuse the same ID for if already generated for a given context/session, else create a new ID and reuse it in every subsequent request",
"required": true
}
]