What is the main goal of this skill?

The primary goal is to create a structured set of questions focused on understanding how a codebase currently works, serving as a foundation for deeper technical research by other agents.

Does this skill suggest code changes?

No, it is strictly designed to focus on the current state of the code to avoid implementation bias and ensure a thorough understanding before any changes are planned.

Which tools does it use to gather information?

It leverages specialized agents like codebase-locator, codebase-analyzer, and codebase-pattern-finder to identify files, trace data flows, and find conventions within your project.

Where are the research questions saved?

Questions are saved as Markdown files within task-specific directories in the 'rpi/tasks/' folder, utilizing date-stamped filenames for easy tracking.

Can I use this for external documentation research?

Yes, it includes a web-search-researcher agent specifically for looking up external SDKs, library documentation, or industry best practices when internal codebase context is insufficient.

Codebase Research Orchestrator

Name: Codebase Research Orchestrator
Author: mshuffett

bymshuffett

0•

Productivity & Workflow

Generates targeted research questions to help AI agents understand the current state and architecture of a codebase before implementation.

This skill acts as a research orchestrator that helps users formulate a comprehensive set of discovery questions about an existing codebase. By focusing exclusively on current implementation details, patterns, and dependencies, it prepares the ground for other agents to execute deep-dive research without introducing premature implementation bias. It integrates with specialized tools like codebase analyzers and pattern finders to ensure every technical nuance is surfaced and documented in a structured, task-specific format, making it ideal for complex onboarding or architectural audits.

Key Features

01Automates the creation of task-specific research documentation and GitHub permalinks

02Generates up to 8 structured research questions based on tasks, specs, or tickets

03Focuses strictly on the current codebase state to prevent implementation leakage

040 GitHub stars

05Follows standardized output templates for consistency across research efforts

06Integrates with specialized agents for code localization and pattern analysis

Use Cases

01Preparing an AI agent to work on a complex legacy system by first mapping its current architecture

02Systematically documenting existing implementation patterns before beginning a major refactoring task

03Mapping out cross-package dependencies in monorepos to conduct an accurate impact analysis

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mshuffett/dotfiles create-research-questions

For use in Claude.ai and ChatGPT

Download Skill