research-agent

Explore codebases read-only to identify architecture and relevant files.

Updated Dec 3, 2025
One-click install
npx skills add https://github.com/Unicorn/Radium --skill research-agent-unicorn
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: research-agent
Source: https://github.com/Unicorn/Radium/tree/main/skills/specialized/research-agent
Command: npx skills add https://github.com/Unicorn/Radium --skill research-agent-unicorn

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Enables safe, read-only exploration of codebases and documentation to quickly understand architecture, locate relevant files, and gather evidence for audits or research without risking accidental modifications.

Core Features & Use Cases

  • Read-only Exploration: Navigate repositories with file reads, directory listings, grep, lint checks, and semantic code search to extract relevant information.
  • Structured Analysis Protocol: Enforces mandatory multi-phase analysis (foundation reads, semantic search, targeted deep dive, synthesis, and introspection) to ensure thorough, reproducible investigation.
  • Use Case: Ideal for onboarding, architecture reviews, security audits, and research tasks where producing file-path-referenced findings and recommendations is required.

Quick Start

Use the research-agent to analyze the repository and produce a summarized architecture report with key files, code patterns, and suggested next steps.

Frequently Asked Questions about research-agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I explore a codebase safely without accidentally modifying files?

To explore a codebase safely, use read-only operations like file reads, directory listings, grep, lint checks, and semantic code search. This approach ensures you can identify architecture and key modules without risking accidental modifications to the repository.

How does semantic code search work for finding implementation patterns?

Semantic code search locates relevant files by matching conceptual queries rather than exact text strings. It enables codebase analysis to surface implementation patterns by reading targeted files and extracting summaries based on the semantic relationships between code elements.

Can I use this approach for security audits and architecture reviews?

Yes, read-only codebase analysis is ideal for security audits, architecture reviews, and onboarding. It locates relevant files, extracts summaries, and produces file-path-referenced findings to gather evidence without modifying the software repository under investigation.

What is the best way to analyze a repository and produce an architecture report?

The best way is to follow a structured multi-phase analysis protocol. Execute foundation reads, semantic search, targeted deep dives, and synthesis to produce a summarized architecture report with key files, code patterns, and suggested next steps.