Skip to content

List tool calls for an execution

GET
/api/tool-calls

Returns all MCP tool calls made during a specific agent execution, ordered chronologically. Each tool call shows what tool was invoked, what arguments were passed, what result was returned, and how long it took. Use this for debugging agent behavior and auditing MCP server interactions.

executionId
required
string format: uuid

The agent execution to list tool calls for.

Tool calls made during the execution.

List of tool calls made during a specific agent execution, ordered chronologically.

Array<object>

A record of a single MCP tool call made by an agent during execution. Tool calls show exactly what data the agent retrieved or what action it took on an external system.

object
id
required

Unique identifier for this tool call.

string format: uuid
Example
tc1a2b3c-5678-9abc-def0-1234567890ab
executionId
required

The agent execution this tool call belongs to.

string format: uuid
Example
ex1a2b3c-5678-9abc-def0-1234567890ab
toolName
required

Name of the MCP tool that was called.

string
Example
get_shipment_status
input

The arguments passed to the tool. Structure varies by tool.

Example
{
"order_id": "4521"
}
output

The result returned by the tool. Structure varies by tool.

Example
{
"status": "in_transit",
"carrier": "DHL",
"tracking_number": "1Z999AA10123456784",
"estimated_delivery": "2026-03-27"
}
status
required

Outcome of the tool call. ‘success’ means the tool returned a result. ‘error’ means the tool returned an error. ‘timeout’ means the tool did not respond in time.

string
Example
success
durationMs

How long the tool call took in milliseconds.

integer | null
Example
245
createdAt
required

Timestamp when the tool call was initiated.

string format: date-time
Example
2026-03-25T11:42:06.000Z

The executionId query parameter is missing.

Standard error response returned by all endpoints on failure.

object
error
required

A human-readable error message describing what went wrong.

string
Example
Conversation not found