> ## Documentation index
> Fetch the complete documentation index at: https://docs.leadmcp.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Exports

Deliver a list as a file.

Starts one file job for a filter and reserves the quota without charging it. The job counts the filter itself. `POST /contacts/count` is free and tells you the price before you pay it. One export runs per account: a second call while one runs returns 409 export.running with the running id. A job delivers at most 50,000 rows. Near a cap the job is truncated to the contact credits left under the cap that runs out first, and a warning gives the reason and the reset time. On the free plan the cap is the lifetime pool: the reason is `monthly_cap`, `resets_at` is `9999-12-31T23:59:59.000Z`, and the hint says the pool does not reset. When the count passes its budget, the server checks for one matching row. A filter that matches nothing still answers 400 export.no_matches, because that check finds no row, and no job starts. A started job then has rows_available null and a count_unavailable warning.

## The file

One export job. The file is UTF-8. A CSV file is RFC 4180: the header row equals the API column names, `id` comes first, a null is empty, a date is RFC 3339 UTC, and the name is `{export_id}.csv.gz`. CSV keeps arrays and objects as JSON text in one cell. A JSONL file uses the same names, one object a line. A JSONL line holds native JSON values, so an array or an object stays an array or an object.

Formats: `csv` or `jsonl`, gzipped. One job writes at most 50,000 rows. A signed link lives 24 hours.

## The job

Statuses: `queued`, `running`, `completed`, `failed`.

Truncation reasons: `daily_cap`, `monthly_cap`, `job_row_limit`, `max_rows`. True when max_rows, the 50,000 row job limit or a contact credit cap made rows_to_export smaller than the rows the filter gives. A per-company cap alone does not set it. The export_truncated warning names the reason and the reset time.