Catch stale GitBook pages every Monday and queue them in Linear

Every Monday at 8am, sweep my GitBook organization for stale documentation and queue up the worst offenders for a refresh.

Walk all spaces with GitBook's List Organization Spaces, then List Pages In Space for each, and surface every page whose last revision is older than 90 days. Skip any space that looks like a changelog (title contains "changelog" or "release notes") and skip archived spaces and archived pages.

For each candidate page, pull the body with Get Page By ID, read what the page actually claims, and form a short opinion on whether it still reflects how the product works today. Look for version numbers that may be outdated, references to old UI elements (buttons, menu names, screen layouts), screenshots that the body implies but cannot be verified, and language that sounds like an older release. Combine the age signal and the content signal into a one-line reason this page looks stale.

Pick the 10 stalest pages (oldest plus highest staleness suspicion). Open one Linear issue using Create Issue titled "Weekly docs cleanup: top 10 stale GitBook pages" with a markdown checklist where each item is the page title, a link to the page, its last-updated date, and the one-line "why I think this is stale" note. Assign the issue to the docs owner and tag it with a docs-cleanup label if one exists.

Then Send a Message via Slack Bot to the #docs channel with a one-paragraph digest: how many spaces were scanned, how many stale pages were found in total, the names of the top 3 worst offenders, and a link to the Linear issue.

Hard rule: never open more than one Linear issue per week. If this Monday's run finds nothing meaningful (no pages older than 90 days), skip both the Linear issue and the Slack post and exit quietly.