PageSift 1.2.2 Release Notes

Release Version: v1.2.2
License: AGPL-3.0-only (Commercial licensing available)

Changes since 1.2.1.

Security and fixes

Crawl and batch job listing, status, errors, streams and cancellation are scoped to the account that created the job. Extract job results are likewise restricted to their owner.
Search with page scraping and extract use the daily page quota. Each page attempt consumes a unit before it is fetched; when the quota is exhausted, further pages are not fetched.
A cached page is checked against robots.txt when the current request respects robots, including recorded redirect URLs.
Expired extract and agent jobs can no longer be retrieved or cancelled.
MCP placeholder outputs match their published output schemas. Nullable fields in the document schema match the serialized result.
Finishing an unreserved itna attempt no longer panics. The session-pool test fixture carries the Retry-After flag into outcome classification.

storeSnapshot: true and removeBase64Images: false now return an unsupported-option error instead of being silently ignored. Change tracking is available through the changeTracking format.
With multiple API keys, an account can no longer access another account's crawl, batch or extract jobs by id.
Engines implementing the Engine trait must implement the budgeted search and extract methods to support those operations for accounts with a daily page quota; the default implementation rejects them.