Scoopkit Client Digest Delivery Setup (Onboarding)

Scoopkit Client Digest Delivery Setup (Onboarding)

1. 01

   Request a magic-link API key at scoopkit.dev and confirm the X-Api-Key header authenticates against api.scoopkit.dev/openapi.json

   Run the key against the OpenAPI spec before touching client data. A key that fails here will fail silently in every downstream webhook, and the client sees an empty digest instead of an error.

2. 02

   Query structured event records and inspect the clustering output for the client's target categories

   Pull event_id, headline, primary_category, article_count, and reported_at. Check that near-duplicate headlines about the same model release collapse into one event with article_count above 1. If the client's niche sits outside the 10 categories and 57 subcategories, flag it now rather than after delivery starts.

3. 03

   Map the client's tracked entities and topics onto Scoopkit search and filter parameters

   Store the mapping in the client workspace file so a second operator can reproduce the same query. This is the step agencies skip, then spend a retainer hour rebuilding when the original operator is on leave.

4. 04

   Choose the delivery mode per client: REST polling, the official CLI via npx scoopkit, or the MCP server for agent workflows

   Polling suits a scheduled digest. The MCP server suits clients whose own agents consume the signal. Pick one per client and write it down; mixing modes across a retainer creates two places to debug.

5. 05

   Register the webhook endpoint and verify a test event lands in the client's CMS or Notion workspace

   Confirm the payload arrives with event_id and reported_at intact. A webhook that fires but drops the timestamp breaks the digest ordering the client sees.

6. 06

   Build the branded digest template and schedule the first send

   Pull deduplicated headlines into the template, not raw article lists. The whole value the client pays for is one event per story, so the template must render article_count rather than repeating near-identical links.

7. 07

   Set the request budget against the plan ceiling and document the archive window

   Free runs 100 req/day with a 45-day archive. Pro runs 10k req/day with a 12-month archive at $49/month. A client needing year-over-year trend comparison cannot sit on Free, and the archive limit is the constraint that forces the upgrade conversation.

8. 08

   Hand off the runbook to the delivery operator with the taxonomy reference and escalation path

   Include the category and subcategory list, the entity mapping, and who to contact when clustering quality degrades. Enterprise limits and archive are custom, so any client at that tier needs the vendor contact recorded in the runbook.