Weekly Linear tickets from Algolia zero-result searches

Every Monday at 9am, pull the top zero-result searches from Algolia for the past 7 days using Algolia's Analytics: Top Searches with No Results. Filter the results to queries whose weekly search count is at or above a configurable minimum threshold (default: 5), then sort the remaining queries by volume descending so the top offenders come first.

For each unique query in the ranked list, first check Linear for an existing open issue in the merchandising team whose title matches the query exactly (using Linear's Search Issues, scoped to the target team and to non-completed workflow states). If a matching open issue is found, skip that query to avoid duplicates.

Otherwise, call Linear's Create Issue in the merchandising team. Use the query text as the issue title. In the description, write a short markdown brief that includes: the query text, the weekly search volume, and a one-line hypothesis on the likely root cause. Pick the hypothesis from three buckets: missing product (the query looks like a product or brand name that we do not carry or have not indexed), missing synonym (a plausible variant of something we do carry, such as a plural, misspelling, or category alias), or content gap (a broader informational query where our catalog copy or landing pages do not yet cover the topic). Note that the hypothesis is a starting point for the merchandiser to confirm.

Expose two knobs the user can tune without editing the workflow: threshold (the minimum weekly search volume before a query becomes a ticket, default 5) and teamId (the Linear team the tickets are filed against, defaulting to the merchandising team). Log a brief summary at the end of each run listing how many queries were considered, how many tickets were created, and how many were skipped as duplicates.