Common Crawl WARC Repackager

Repackage the Web page records for your hostnames/domains across selected Common Crawl archives into a new WARC archive in your Hugging Face bucket. Compute runs on HF Jobs data is read from the Common Crawl HF bucket.

Jobs are available to any user or organization with a positive credit balance.

1. Filter target

Registered domains

Matches the domain and all its subdomains (url_host_registered_domain).

Exact hostnames

Matches exact hosts only (url_host_name). Combine with domains.

Content languages (optional)

ISO-639-3 codes (content_languages). Keeps records whose primary detected language is one of these. Leave empty for all languages.

2. Target crawls

Crawls to scan

Selecting > 10 crawls makes the index scan large.

CC-MAIN-2026-25 CC-MAIN-2026-21 CC-MAIN-2026-17

3. Output target

Pick one of your existing buckets (the dropdown lists buckets you can write to). Create a new one at https://huggingface.co/new-bucket, then click Refresh. Dataset repos are not supported as a target.

Target bucket

Buckets you can write to (yours + write-access orgs).

Path within bucket

Sub-folder for the output; created if missing. Leave blank for the bucket root.

Output WARC name

Hardware flavor

4. Estimate & launch

Job logs