What problem does it solve?
Provides a controlled, evidence-driven research phase that prevents premature coding and ensures a comprehensive understanding of a repository before planning or implementation begins. It centralizes architecture, patterns, data layer, API, infrastructure, and dependency analysis into a single reproducible document while enforcing read-only constraints.
Core Features & Use Cases
- Sessioned State Loading: Resolves session ID, loads session state, and enforces phase correctness for repeatable research runs.
- Incremental & Scoped Analysis: Supports --scope and --incremental modes to re-analyze only specified areas or diffs.
- Solo and Team Modes: Runs as a solo investigator or spawns coordinated agents to produce partitioned analyses that are later synthesized.
- Cross-Plugin Context: Reads auxiliary reports like harnessability and evolve-insights to enrich the research context.
- Topology Detection & Health Checks: Invokes topology detector and health-check hooks to surface architecture and fitness signals.
- Non-destructive Output: Writes a structured workdir research.md with RF/RA tagging and a research quality checklist while forbidding source modifications.
Quick Start
Start phase 1 research on the repository to produce a complete research.md in the project work directory covering architecture, patterns, data, API, infrastructure, and dependency risk analysis.
