---
title: "Batch settled"
method: POST
path: "batchSettled"
tags: ["Batches"]
---

# Batch settled

`POST batchSettled` (webhook)

Sent to a batch's `webhookUrl` once it reaches a final status, whether it completed, was cancelled, or failed. Delivery is best effort and at most once: one attempt, a 10 second timeout, no retries, and any response status is treated as acknowledged. Polling GET /batch/{batch_id} remains the authoritative completion signal. The embedded batch carries signed result URLs that expire, so download them promptly or re-read the batch.

## Headers

- `User-Agent` string, required
- `X-Context-Event` 'batch.completed' | 'batch.cancelled' | 'batch.failed', required
- `X-Context-Id` string, required
- `X-Context-Signature` string

## Payload

- BatchSettledWebhookPayload — Payload sent when a batch reaches a final status.
  - `event` 'batch.completed' | 'batch.cancelled' | 'batch.failed', required — Which final status the batch reached. Mirrors the `X-Context-Event` header.
  - `id` string, required — Unique event ID. Mirrors the `X-Context-Id` header.
  - `created_at` string, date-time, required — When the delivery was generated.
  - `data` object, required
    - `batch` Batch, required — An asynchronous web scraping job.
      - `id` string, required — Batch ID used to retrieve or cancel the job.
      - `status` 'queued' | 'running' | 'cancelling' | 'completed' | 'cancelled' | 'failed', required — Current state. `completed`, `cancelled`, and `failed` are final.
      - `mode` 'scrape' | 'crawl', required — How pages were selected. Matches `input.mode` on the submit request.
      - `format` 'markdown' | 'html', required — What each page is returned as. Matches `input.data.format` on the submit request.
      - `tags` string[], required — Tags stored on the batch at submission.
      - `crawl` BatchNullableCrawlControls, required — The crawl controls as submitted, so the limits requested can be compared against what the crawl reached.
        - `source` union, required — Where the crawl started.
          - object — The crawl discovered pages by following links from one URL.
            - `type` 'start_url', required
            - `url` string, required — Page the crawl started from.
          - object — The crawl scraped the pages listed in the domain's sitemap.
            - `type` 'sitemap', required
            - `domain` string, required — Domain whose sitemap supplied the pages.
        - `max_pages` integer, required — The `maxUrls` submitted with the crawl. A sitemap crawl scrapes only the URLs its sitemap actually lists, up to this many, so `input.reserved` is often lower.
        - `max_depth` integer, nullable, required — Link depth limit. Always 0 for a sitemap crawl, which never follows links off its URLs; null when a `start_url` crawl set no limit.
        - `follow_subdomains` boolean, required — Whether links to subdomains were followed. Always false for a sitemap crawl.
        - `url_pattern` string, nullable, required — RE2 pattern URLs had to match to be crawled. Null when the crawl set none.
      - `input` BatchIntake, required — What submission took in, and what it charged for.
        - `reserved` integer, required — Pages credits were reserved for. Everything else — progress, the refund, the completion percentage — is measured against this.
        - `reserved_is_ceiling` boolean, required — Whether `reserved` is an upper bound the batch may finish under. True only for a crawl that follows links, whose reachable page count is unknowable until it runs. False for a scrape and for a sitemap crawl, where `reserved` is an exact page count.
        - `submitted` integer, nullable, required — URLs in the list you sent, before validation and de-duplication. Null for a crawl, which is given a source rather than a list.
        - `duplicates` integer, required — URLs dropped before reserving because another entry resolved to the same page. Non-zero for sitemap crawls too, whose sitemaps routinely list a page more than once.
        - `invalid` integer, nullable, required — URLs from your list rejected as unusable; the same ones are itemised in `invalid_urls` at submission. Null for a crawl — a crawl that resolves no usable page is rejected outright with a 400 rather than accepted with an empty list.
      - `progress` object, required — Pages attempted so far. Use `status` to check completion.
        - `succeeded` integer, required — Pages scraped successfully.
        - `failed` integer, required — Pages that could not be scraped.
        - `pending` integer, required — Reserved pages not yet attempted. A cancelled batch keeps reporting the URLs it never reached; a crawl whose `input.reserved_is_ceiling` is true reports 0 once final, because its unspent budget was never real pages.
      - `credits` object, required — What this batch has done to your credit balance.
        - `reserved` integer, required — Credits debited from your balance the moment the batch was accepted. This is a charge, not a forecast — the whole amount leaves the balance up front.
        - `refunded` integer, required — Credits returned for pages that did not succeed. Stays 0 until the batch reaches a final status, then settles in one movement.
        - `ocr_charged` integer, required — Credits charged for PDF pages recovered by OCR (pdf.ocr=true), 1 per recovered page, on top of `reserved`. Stays 0 until the batch settles.
        - `net` integer, required — `reserved` minus `refunded` plus `ocr_charged` — what the batch has cost so far. Equal to `reserved` until the batch settles.
      - `timing` object, required
        - `created_at` string, required — When the batch was created.
        - `started_at` string, nullable, required — When processing started. Null while queued.
        - `completed_at` string, nullable, required — When processing finished. Null while active.
      - `page_errors` BatchPageErrorCount[], required — Individual page failures grouped by error code, sorted by count. Unrelated to `failure`, which is the batch itself failing.
        - `code` string, required — Error code for these failures.
        - `count` integer, required — Pages that failed with this code.
      - `failure` BatchNullableFailure, required — A failure of the batch as a whole, distinct from the per-page failures in `page_errors`.
        - `code` string, required — Why the batch itself stopped.
        - `message` string, required — Human-readable explanation.
      - `results` object, nullable, required — Download links, available once the batch reaches a final status and null before then. GET /batch/{batch_id}/results serves the same records as paginated JSON.
        - `expires_at` string, required — When the download URLs expire.
        - `files` object[], required — Result files. Order is not guaranteed.
          - `url` string, required — Temporary URL for a gzipped NDJSON file.
          - `items` integer, required — Results in this file.
          - `bytes` integer, required — Compressed file size in bytes.

## Acknowledgement `2XX`

Webhook accepted.

---

[API](https://skmtc.net/context/apis/context-dev.md) · [All operations](https://skmtc.net/context/apis/context-dev/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/context/context-dev/versions/3c6a01ba967f/schema)