Skip to main content
GET
Export Test Run CSV

Overview

Dispositions are enabled per organization. If your organization does not have access, these endpoints return 404.
Streams the whole test run as a CSV file, one row per call and value, with the engine’s result, its evidence, and the latest correction saved for that cell. This is the same data as Get Test Run Rows without paging. Unlike the other endpoints here, the success response is CSV, not JSON; errors still use the JSON envelope.

Headers

string
required
Your API key for authentication.

Path Parameters

string
required
The agent’s unique identifier. Must be a UUID; otherwise returns 400 with the message agentId must be a valid UUID. Returns 404 Agent not found if the agent is not in your organization.
string
required
The disposition’s unique identifier. Must be a UUID; otherwise returns 400 with the message dispositionId must be a valid UUID. Returns 404 Disposition not found if it does not belong to this agent.
string
required
The test run’s unique identifier. Must be a UUID; otherwise returns 400 with the message runId must be a valid UUID. Returns 404 Test run not found if it does not belong to this disposition.

Response

Returns 200 with Content-Type: text/csv; charset=utf-8 and Content-Disposition: attachment; filename="disposition-run-{run_id}.csv". The body starts with a UTF-8 byte order mark, rows end in CRLF, and cells containing commas, quotes, or line breaks are quoted with doubled inner quotes. A text cell that begins with =, +, @, or - and is not a number is prefixed with an apostrophe so spreadsheets do not treat it as a formula. Rows are sorted by call_id, then by definition order of the values, so a run with 50 calls and 3 values yields 150 rows. Only the latest correction for each call and value appears; the export reads at most the 25,000 most recent correction revisions, which covers a full 250-call, 50-value run twice over.
string
Identifier of this call’s execution within the run. It matches testRunCaseId on correction objects.
string
The call.
string
Execution status for the call: pending, running, complete, failed, or cancelled.
string
When the call’s execution failed as a whole, one of the run error codes (for example disposition_run_failed); otherwise empty.
string
Human-readable message for run_error_code, or empty.
string
The definition value this row is about.
string
The value’s key.
string
The value’s label.
string
The engine’s result state: produced, no_value, error, or missing when the call produced no result for this value.
string
The engine’s produced value, JSON-encoded (so a string appears in quotes, a boolean as true or false). Empty when there is no value.
string
JSON-encoded array of evidence quotes, each with source, speaker, start_ms, end_ms, and text. Empty when none.
string
State of the latest correction, produced or no_value, or empty if the cell was never corrected.
string
The corrected value, JSON-encoded, or empty.
string
Revision number of the latest correction, or empty.
string
The explanation supplied with the latest correction, or empty.
string
When original_state is error, the value-level error code (for example judge_failed); otherwise empty.
string
Message for value_error_code, always This value could not be produced. when set; otherwise empty.

Docs for agents: llms.txt