Open in ChatGPTOpen in Claude

Reference tools

The MCP server provides foundational tools for understanding and querying your data effectively.

Explore DataPrime commands and functions

Use the list_dataprime_commands tool to browse the available DataPrime commands and functions, and the show_dataprime_command_details tool to see the syntax, description, and category for a specific one. These tools help AI agents construct correct queries with DataPrime, Coralogix's structured query language for logs, traces, and metrics. They are based on the DataPrime Reference.

Explore the relevant commands before running queries with query_dataprime and query_promql_range to ensure correct syntax and interpretation.

list_dataprime_commands parameters:

Name	Required	Type	Description
kind	No	enum	One of all (default), commands, or functions.
name	No	string	Case-insensitive substring to filter the catalog by command or function name.

show_dataprime_command_details parameters:

Name	Required	Type	Description
kind	Yes	enum	One of all, commands, or functions.
name	Yes	string	The exact command or function name to show details for.

Discover fields with search_for_fields

Use the search_for_fields tool to discover the log and span field paths (keypaths: $d, $l, $m) in your data before writing complex queries. Search by field description, by an example value, or by a batch of both—up to five searches per call.

Example:

{
  "search_items": [
    { "description": "http status code" },
    { "value": "payment-service" }
  ],
  "source": "logs"
}

Parameters:

Name	Required	Type	Description
search_items	Yes	array	Up to five search items. Each item has either a description (semantic search over field descriptions) or a value (case-insensitive substring match on example values), but not both.
source	No	enum	One of logs (default) or spans.

Helpful hints:

• Use description searches to find fields by meaning, and value searches to find which field contains a known value.
• To discover metrics instead of log or span fields, use search_relevant_metrics.

Discover datasets

Beyond source logs and source spans, you can query named datasets—both Coralogix system datasets (system/...) and your own user-defined datasets—with DataPrime. Use these tools to discover which named sources are available and to inspect a system dataset's schema before querying it. Both tools are available on MCP server v2.

Use the list_datasets tool to list the dataset sources available for DataPrime queries. This tool takes no parameters and returns a JSON string with:

• alwaysAvailableSources — the built-in sources source logs and source spans.
• datasetCounts — per-type returned and total counts for system and user datasets.
• truncated — true when more datasets exist than were returned (each type is capped at 50 results).
• datasets — an array of dataset entries. Each has a type (system or user), a source you can query with source <name>, and type-specific metadata such as description and queryEnabled (system) or writeEnabled (user).

Use the get_system_dataset_info tool after list_datasets when you want to understand what a specific system dataset contains before querying it.

Parameters:

Name	Required	Type	Description
dataset	Yes	string	A single system dataset path, such as system/engine.queries.

It returns a JSON string with the dataset source, its description, optional additional_context_markdown guidance, and the dataset schema.

Get the current time with get_datetime

Use the get_datetime tool to retrieve the current date and time. This is useful for constructing time-bounded queries when you need a reference to the present moment.