domain-research

Subdomain discovery uses parallel four-agent research to identify candidate subdomains within a target domain, reducing bias and broadening coverage. It classifies subdomains by task type and complexity, then maps them to initial pipeline concepts for further development.

Domain decomposition breaks a target domain into manageable subdomains to drive a self-improving pipeline generator. You need it when classifying subdomains by reuse potential and mapping them to step-menu chains to produce a Component Manifest for the scaffolder.

To start mapping subdomains, provide a target domain name and a run ID to initiate Phase 1 research. The process generates phase artifacts, specifically content.md and manifest.json files, across phases 1 through 3 to document the pipeline concepts.

Yes, you can integrate existing repository inventory and domain references to guide subdomain classification. The research agents reference a task-type guide to classify discovered subdomains by task type, complexity, and reuse potential before mapping them.

The best way to reduce bias during subdomain discovery is using parallel research with four independent agents. This multi-agent approach broadens the identification of candidate subdomains within a target domain before classification and pipeline mapping.

Domain decomposition produces phase artifacts including content.md and manifest.json files for phases 1 through 3. The Component Manifest specifically maps classified subdomains to step-menu chains, providing structured output directly usable by the scaffolder.