Weekly Context7 docs usage report in Sheets and Slack

Every Monday at 9am, run a scheduled workflow that reports on how AI coding assistants consumed the documentation we publish to Context7 over the previous 7 days. This is a fixed report with fixed fields and fixed destinations, so build it as a deterministic workflow with no reasoning step.

Keep the list of library IDs we own as a configurable input at the top of the workflow so we can add or remove libraries without rebuilding it. For each library in that list, call Get Library Usage Metrics in Context7 with the days parameter set to 7. From each response, pull the request counters, the MCP client breakdown, the topic distribution and the country distribution. Note that this operation requires our Context7 API key to belong to a member of the teamspace or team project that owns the library, so it only works for libraries we publish ourselves.

Before writing anything new, use Get Values in Google Sheets to read the existing rows on the tracking tab and find the most recent prior row for each library. Use that row's total request count to compute the week over week change, as both an absolute difference and a percentage. If there is no prior row for a library, treat this run as the baseline and record the change as not applicable rather than as zero.

Then use Append Values in Google Sheets to append one row per library per week to the tracking tab. Each row should carry: the week start date, the week end date, the library ID, total requests for the window, the absolute and percentage change versus the prior week, the per client request counts for the MCP clients returned (Cursor, Claude Code, Windsurf and any others), the top topics with their request counts, and the top countries with their request counts. Serialize the client, topic and country breakdowns into single cells as compact name equals count lists so the row stays one row wide and the column layout never shifts between weeks. Always append a row for every tracked library, even when a library returned no traffic, recording zeros so the history has no gaps.

Finally, use Send a Message in Slack to post the weekly summary to our developer relations channel. The message should lead with the reporting window and the combined request total across all libraries with its week over week change, then list each library on its own line with its total and its change versus last week, then show the aggregate client breakdown so we can see which coding assistants are hitting our docs hardest, and close with the most requested topics across all libraries. Sort libraries by request volume descending and mark clear risers and fallers so the trend is readable at a glance.

If Context7 returns an error for one library, skip that library, still report the others, and note the failure at the end of the Slack message rather than aborting the whole run.