The public production MCP endpoint is live at https://mcp.everyn.ai/mcp. Configure a host with that Streamable HTTP URL and complete WorkOS OAuth.

Do not stop at terminal Run status. When get_run says next_action is export_and_analyze, call export_run, download the CSV, and analyze that file for the user.

Do not ask for a second approval to export or analyze an already authorized Run. Host-native confirmation may still appear for the original cost-bearing start.

Export

export_run creates or reconnects to the complete terminal CSV export and returns a refreshable HTTPS delivery descriptor. The large CSV never enters MCP content.

If the descriptor expired, call export_run again with the same idempotency key to refresh delivery. Do not fetch arbitrary URLs.

not_ready means the Run is not yet exportable. Keep waiting with get_run.

Complete-results CSV

The public projection is complete_results. It does not include retry-graph or internal lineage fields.

Reserved columns:

_everyn_dataset_row_id
_everyn_result_state
_everyn_failure_code
_everyn_failure_message
_everyn_row_kind

If a user-authored column equals a reserved name, Everyn keeps the user column and suffixes the reserved header with _reserved.

Both add_columns and create_rows preserve source-row identity through _everyn_dataset_row_id.

Preview versus export

get_run_results is a bounded preview for inspection. It is not a substitute for the complete CSV. Analyze the downloaded file, not the preview rows, when reporting full-Run outcomes.