Phase Research Analyst

Overview

This agent acts as a dedicated technical researcher, designed to answer the critical question: "What do I need to know to plan this phase well?" It synthesizes information from provided context files and general knowledge into a single, structured RESEARCH.md document.

The output is specifically formatted for consumption by a subsequent planning agent, ensuring all factual claims are properly attributed with provenance tags (e.g., [VERIFIED: npm registry], [CITED: docs...]).

Capabilities

• Contextual Deep Dive: Reads and synthesizes information from multiple source files provided in the prompt.
• Domain Investigation: Identifies standard technology stacks, architectural patterns, and common pitfalls for a given technical phase.
• Provenance Tracking: Tags every factual claim with its source of truth (Verified, Cited, or Assumed) to maintain high accuracy and flag areas needing user confirmation.
• Structured Output Generation: Produces a clean RESEARCH.md file adhering to expected planning structures.

Example Use Cases

1. New Feature Implementation: When starting a complex feature (e.g., implementing OAuth 2.0), use this agent to research the required standards, best practices, and necessary dependencies before writing any code or detailed plan.
2. Technology Stack Evaluation: If the team is debating between two frameworks for a new service, feed in documentation snippets for both and ask the agent to compare them based on industry standards and documented pitfalls.
3. Compliance Review: For regulated domains, use it to cross-reference project guidelines (CLAUDE.md) against external best practices, ensuring all security or compliance requirements are noted with appropriate confidence levels.