Platform limits

Limit	Value
Browser concurrency	120 per account by default
Per-host politeness	1 second, 2 concurrent by default
URLs per batch	100
Crawl pages	5,000
Crawl depth	10
Crawl concurrency	10 pages
Crawl duration	6 hours
Screenshot resolution	3,840 × 2,160
Render timeout	60 seconds
Stored job result	approximately 1.5 MB
Ordinary cache entry	24 MB
Inline design JSON	2 MiB
AI projection	16 KiB
AI provider response body	512 KiB
AI provider deadline	30 seconds
Design stylesheets	60 sheets, 40,000 rules, 2 MB total

Limits are validation ceilings, not performance targets. Full-page screenshots, script-heavy sites, large DOMs, and stylesheet-heavy design extraction can take longer than simple snapshots.

Pagination and truncation

Crawl pages use a cursor. Send the returned cursor unchanged on the next request rather than constructing offsets. Async job results must fit in D1. When a result exceeds about 1.5 MB, Scrinly removes large optional fields in a deterministic order and records their names in result.truncated. Prefer object storage when you need complete large artifacts.

Getting started

Core workflows

Scale and delivery

Reliability

​Pagination and truncation

Pagination and truncation