Skip to main content
Human guide: task steps and agent prompt → Read Deepline’s global funding feed and keep progress in your own system. This public pull feed requires no Monitor deployment. Each request is metered, including later cursor pages.

1. Read the first page

Inspect the current contract and price:
This request returns up to the 100 most recent discoveries:
For a later poll, include your saved since timestamp in the first request. since filters discovery time, not the funding announcement date. The feed has no company, industry, round-stage, or source filter.

2. Consume every cursor page

Persist each successful page before continuing. When data.has_more is true, pass that response’s exact data.next_cursor as cursor:
Replace the placeholder with the prior page’s cursor. Keep it unchanged. Continue until has_more is false. Retain failed pages as unfinished work; a successful first page does not complete the read.

3. Advance the checkpoint after processing

Store the original discovery-time boundary, successful pages, and next cursor. Use update_id to deduplicate discoveries across repeated or overlapping reads. since is inclusive, so the boundary can return an already processed update. Advance the saved discovery-time checkpoint only after every page has been stored and processed. Preserve it when any page fails. Funding announcements and response times are not substitutes for the discovery boundary. The feed creates neither a schedule nor a Monitor. For recurrence, own the checkpoint and retry policy in a Play or scheduler. Choose the interval for your review needs; no publication cadence is promised. Use Review account signals to qualify the retrieved events before account actions.

Call the API

Use a workspace-scoped API key from a trusted server or worker. Do not expose the key in browser code.
For the next page, persist the prior response and build the request from its cursor. Only run this when the prior response has data.has_more: true.