gsd-research-phase

Automating phase research involves spawning a dedicated researcher agent to validate feasibility against your roadmap. This process checks existing research to avoid duplication and aligns with phase goals before generating a standardized RESEARCH.md document for decision gates.

Structured context loading for isolated phase research requires providing specific inputs like phase_number and phase_name. This ensures the automated researcher agent gathers accurate context, aligns with the roadmap, and properly outputs architecture and outcome sections.

Generating a standardized RESEARCH.md requires invoking an automated phase research workflow with a specific phase number. The system then produces the file with prescribed sections for architecture, outcomes, and next steps while running built-in quality checks.

Yes, an existing roadmap is required before starting automated phase research. The research process validates phase feasibility by checking alignment against the roadmap, ensuring that isolated implementation planning integrates smoothly into the broader project goals.

Checking existing research to avoid duplication is a core feature of automated phase planning. The workflow explicitly evaluates prior research context before proceeding, ensuring new analysis builds on existing findings rather than repeating previous work.

Researching a phase in isolation focuses strictly on validating feasibility before integration. It does not execute the implementation itself; instead, it produces a RESEARCH.md decision gate document that must be reviewed before proceeding to actual planning and integration steps.