Codebase Onboarding

Codebase Onboarding Skill

A comprehensive skill to help new developers understand large codebases quickly by generating architecture overviews, identifying entry points, tracing data flows, and creating navigable codebase maps.

Quick Start

When a user asks to understand a codebase:

Onboarding Progress:
- [ ] Step 1: Analyze project structure
- [ ] Step 2: Identify entry points
- [ ] Step 3: Map key components
- [ ] Step 4: Trace data flows
- [ ] Step 5: Document patterns
- [ ] Step 6: Generate onboarding guide

Workflow

Step 1: Analyze Project Structure

Run the structure analyzer to understand the project layout:

python .claude/skills/codebase-onboarding/scripts/analyze_structure.py .

This will output:

• Project type (Node.js, Python, Go, etc.)
• Directory structure with descriptions
• Key configuration files found
• Detected frameworks and libraries

Step 2: Identify Entry Points

Find where the application starts and its main interfaces:

python .claude/skills/codebase-onboarding/scripts/find_entry_points.py .

Entry points include:

• Main application files (index.js, main.py, main.go)
• API route handlers
• CLI entry points
• Event handlers and listeners

Step 3: Map Key Components

Read and analyze the core components:

1. Models/Types - Data structures used throughout
2. Services - Business logic implementations
3. Controllers/Handlers - Request handling
4. Utilities - Shared helper functions
5. Configuration - App settings and constants

Use Glob and Read tools to explore these directories.

Step 4: Trace Data Flows

For each major feature, trace how data moves:

python .claude/skills/codebase-onboarding/scripts/trace_data_flow.py . --feature "user authentication"

Document:

• Input sources (API, CLI, events)
• Processing steps
• Data transformations
• Output destinations (DB, API, files)

Step 5: Document Patterns

Identify recurring patterns in the codebase. See [PATTERNS.md](PATTERNS.md) for common patterns to look for.

Key patterns to identify:

• Architectural patterns (MVC, Clean Architecture, etc.)
• Error handling approaches
• Logging conventions
• Testing strategies
• Naming conventions

Step 6: Generate Onboarding Guide

Create a comprehensive onboarding document:

python .claude/skills/codebase-onboarding/scripts/generate_map.py . --output ONBOARDING.md

The guide should include:

• Architecture diagram (Mermaid)
• Component overview
• Data flow diagrams
• "Where to find X" quick reference
• Common tasks guide

Output Format

Generate documentation in this structure:

# [Project Name] - Developer Onboarding Guide

## Architecture Overview

[Mermaid diagram showing major components]

## Project Structure

[Directory tree with descriptions]

## Key Components

[Table of important files and their purposes]

## Data Flows

[Diagrams showing how data moves through the system]

## Common Tasks

- How to add a new API endpoint
- How to add a new feature
- How to run tests
- How to deploy

## Patterns & Conventions

[Coding standards used in this project]

Examples

Example 1: Node.js Express API

User: "Help me understand this codebase"

Steps:

1. Run analyze_structure.py → Detects Node.js/Express
2. Run find_entry_points.py → Finds src/index.js, route files
3. Read package.json, tsconfig.json for dependencies
4. Map: routes/ → controllers/ → services/ → models/
5. Generate architecture diagram
6. Create ONBOARDING.md with quick reference

Example 2: Python Django Project

User: "I'm new to this project, give me an overview"

Steps:

1. Detect Django from manage.py, settings.py
2. Find apps in INSTALLED_APPS
3. Map: urls.py → views.py → models.py
4. Trace request flow through middleware
5. Document ORM patterns and migrations
6. Generate Django-specific onboarding guide

Tips for Success

1. Start high-level - Don't dive into implementation details initially
2. Follow imports - Trace import chains to understand dependencies
3. Read tests - Tests often reveal intended behavior
4. Check docs/ - Existing documentation is valuable context
5. Look for README - Project README often explains structure

Related Skills

• For API documentation: see API docs skill
• For dependency analysis: see dependency audit skill
• For code quality: see code review skill

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: dcs-soni
• Source: dcs-soni/skills
• License: MIT

Install and usage instructions live in the source repository linked above.