MCP Tools Reference

Through MCP, agents can use the following tools to interact with the Context File System.

Availability for each tool is controlled by your File Level Security (FLS) configuration.

Tools overview

Read-only tools

get_data_schema

Returns the current data structure so the agent can understand which fields exist and what their types are. Agents usually call this before querying.

Input parameters: None

Example response:

{
  "schema": {
    "type": "array",
    "items": {
      "type": "object",
      "properties": {
        "id": { "type": "string" },
        "name": { "type": "string" },
        "price": { "type": "number" },
        "status": { "type": "string", "enum": ["active", "inactive"] }
      }
    }
  }
}

get_all_data

Returns the full dataset with no filtering. This is best for smaller datasets.

Input parameters: None

Example response:

[
  { "id": "1", "name": "Widget Pro", "price": 99.99 },
  { "id": "2", "name": "Widget Mini", "price": 59.99 }
]

query_data

Runs a conditional query with support for filters, sorting, and limits.

Input parameters:

Filter syntax:

// Equals
{ "status": { "$eq": "active" } }
 
// Greater than
{ "price": { "$gt": 50 } }
 
// Included in an array
{ "category": { "$in": ["electronics", "accessories"] } }
 
// Combined conditions, AND
{
  "$and": [
    { "status": { "$eq": "active" } },
    { "price": { "$lt": 100 } }
  ]
}
 
// Combined conditions, OR
{
  "$or": [
    { "priority": { "$eq": "urgent" } },
    { "priority": { "$eq": "high" } }
  ]
}

Supported operators: $eq, $ne, $gt, $gte, $lt, $lte, $in, $nin

preview

Quickly previews the first few records so the agent can understand the shape of the data without pulling everything.

Input parameters:

select

Returns only specified fields, which reduces token usage when the dataset contains many fields but the agent only needs a few of them.

Input parameters:

Example:

{
  "fields": ["id", "name", "price"],
  "filter": { "status": { "$eq": "active" } }
}

Write tools

Write tools must be explicitly enabled in FLS configuration before they can be used.

create

Creates a new item in the dataset.

Input parameters:

Example:

{
  "data": {
    "name": "New Widget",
    "price": 79.99,
    "status": "active"
  }
}

update

Modifies fields on existing items.

Input parameters:

Example:

{
  "filter": { "id": { "$eq": "123" } },
  "data": { "price": 89.99, "status": "active" }
}

delete

Deletes items matching the filter. This action cannot be undone, so authorize it carefully.

Input parameters:

Example:

{
  "filter": { "status": { "$eq": "archived" } }
}

Customizing tool names

Default tool names such as create_element can be customized in MCP configuration so the agent gets clearer context:

{
  "tools_definition": {
    "create": {
      "name": "add_product",
      "description": "Add a new product to the catalog"
    },
    "query_data": {
      "name": "search_products",
      "description": "Search products by condition"
    }
  }
}