Crawl
Turns the scattered public record into a maintained hub.
Crawl is the agentic public-source engine. Where Frame and Forge create first-party supply from the field, Crawl reaches the public record: the registries, portals, reports, and datasets that already exist but sit scattered, in incompatible shapes, at unknown freshness, behind no single front door. It runs a four-move loop, scan, verify, organise, maintain, under human curation.
It produces two things from one engine. A human-facing Hub: a curated, source-rated, maintained reference anyone can use, from local stakeholders and researchers to app users who need context alongside their own records. And a supply stream: the same organised public data, which becomes tradable once Scribe renders it into conforming records and Spine exchanges it, exactly as any external database or data dump does.
The method is the most field-proven in the suite. Hand-run crawls already maintain live reference hubs for cobalt, extractives transparency, artisanal mining, and Global-South climate. What is being built now is the sector-agnostic engine underneath them, configured per sector rather than rewritten.
Where it sits. A public-data supply source. Its Hub output is standalone; its market-facing output flows through Scribe to conform, then Spine to exchange, the same path any external source follows.