Onboarding Claude Code like a new developer: Lessons from 17 years of development

2026-04-28 · Source: Claude Blog · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning · Depth: Intermediate, medium

Summary

Brendan MacLean of the University of Washington's MacCoss Lab successfully applied his 17-year methodology for onboarding new human developers to Claude Code, an AI coding tool, for their 700,000-line C# codebase, Skyline. Skyline, an open-source protein analysis software in development since 2008, benefits from this approach by making its complex, long-lived codebase more manageable. MacLean's strategy involves treating Claude Code like a trainee developer, providing it with structured context and skills in a separate `pwiz-ai` repository. This method enabled the completion of a year-long Files View panel project in two weeks and accelerated development on other features, including automating screenshot reproduction and daily summary emails from nightly test infrastructure. The approach has shifted developers' roles from writing code to instructing Claude Code, leading to increased feature development.

Key takeaway

For AI Engineers managing large, legacy codebases, adopting a structured onboarding process for AI coding tools like Claude Code is crucial. By investing in a dedicated context layer and skill library, you can significantly reduce technical debt and accelerate feature development, transforming AI from a basic code generator into an integrated, knowledgeable team member. This approach allows your team to focus on higher-level problem-solving rather than repetitive coding tasks.

Key insights

Treating AI coding tools like new human developers, with structured onboarding and context, significantly enhances their effectiveness on complex codebases.

Principles

• Context is a project artifact to be versioned and maintained.
• Skills encode domain knowledge for AI agents.
• MCP integrations enable AI access to real-time data.

Method

Onboard AI by providing limited project scope, then iteratively expand context. Store AI context in a separate, versioned repository. Develop a skill library for domain knowledge and use MCP integrations for real-time data access.

In practice

• Create a `CLAUDE.md` for environment setup.
• Develop a `debugging` skill for root cause analysis.
• Automate daily summaries from test infrastructure.

Topics

• Claude Code
• Legacy Code Management
• AI-Assisted Development
• Context Engineering
• Software Automation

Code references

• ProteoWizard/pwiz-ai

Best for: Software Engineer, AI Engineer, Director of AI/ML

Related on AIssential

• Claude Code: The 2026 Zero to Hero Guide — Claude Code transforms engineering by enabling autonomous, agentic code development and system orchestration.
• AI coding tools with organizational context are quietly changing how engineering onboarding works — AI coding tools with organizational context accelerate new engineer onboarding by demonstrating codebase conventions.
• What's going on with Claude Code? — AI models are rapidly transforming software development, shifting human roles from coding to supervision and taste.
• How to use Claude Code to build a board room of world class Product Mentors — Claude Code enables non-engineers to build AI mentor systems for strategic product decision-making.
• zilliztech / claude-context — Claude Context provides AI coding agents with cost-effective, deep codebase context via semantic search and vector databases.
• Eight years of wanting, three months of building with AI — AI excels at tedious coding tasks but struggles with high-level architectural design and subjective problem-solving.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Claude Blog.