Configuration

WYN360 CLI uses a hierarchical configuration system that allows you to customize behavior at multiple levels.

Configuration Hierarchy

Configuration is loaded in this priority order (highest to lowest):

1. Project Configuration (.wyn360.yaml) - Highest priority
2. User Configuration (~/.wyn360/config.yaml)
3. Default Values - Lowest priority

User Configuration

Location

~/.wyn360/config.yaml

Creating User Config

Create your user configuration directory and file:

mkdir -p ~/.wyn360
touch ~/.wyn360/config.yaml

User Config Example

# ~/.wyn360/config.yaml
model: "claude-sonnet-4-20250514"
max_tokens: 4096
temperature: 0.1

# Custom instructions added to every conversation
custom_instructions: |
  Please follow these coding standards:
  - Use type hints for all function parameters and return values
  - Add docstrings to all functions and classes
  - Follow PEP 8 style guidelines
  - Prefer pathlib over os.path for file operations
  - Use f-strings for string formatting

# Browser use settings
browser_use_cache_enabled: true
browser_use_cache_ttl: 1800  # 30 minutes
browser_use_truncate_strategy: "smart"  # smart, beginning, end, middle
browser_use_max_tokens: 10000

# User aliases and shortcuts
aliases:
  review: "Review this code for potential improvements, bugs, and style issues"
  test: "Generate comprehensive pytest tests for this code"
  doc: "Add comprehensive documentation and docstrings to this code"

Project Configuration

Location

.wyn360.yaml  # In your project root directory

Project Config Example

# .wyn360.yaml - Project-specific settings
model: "gemini-2.5-flash"  # Cost-effective for this project
max_tokens: 2048

# Project context helps AI understand your setup
project_context: |
  This is a FastAPI web application with:
  - PostgreSQL database using SQLAlchemy ORM
  - Redis for caching
  - Celery for background tasks
  - Docker for containerization
  - pytest for testing

  Code style preferences:
  - Use async/await for all database operations
  - Prefer Pydantic models for request/response validation
  - Store configuration in environment variables
  - Use dependency injection for database sessions

# Project-specific custom instructions
custom_instructions: |
  For this project:
  - Always use async/await patterns
  - Include proper error handling with custom exceptions
  - Add logging statements for important operations
  - Follow the established project structure in /app

# Browser settings for this project's needs
browser_use_cache_enabled: false  # Always fetch fresh data
browser_use_max_tokens: 15000     # Need more context for documentation

Configuration Options Reference

Core Settings

Option	Type	Default	Description
model	string	Auto-detect	AI model to use
max_tokens	integer	4096	Maximum output tokens
temperature	float	0.1	Model creativity (0.0-1.0)
custom_instructions	string	""	Added to every conversation
project_context	string	""	Project-specific context

Browser Use Settings

Option	Type	Default	Description
browser_use_cache_enabled	boolean	true	Enable website caching
browser_use_cache_ttl	integer	1800	Cache TTL in seconds
browser_use_truncate_strategy	string	"smart"	How to truncate long pages
browser_use_max_tokens	integer	10000	Max tokens for fetched content

Truncate Strategies

• smart - Preserves structure, removes less important content
• beginning - Keep first N tokens
• end - Keep last N tokens
• middle - Keep first and last, remove middle

Environment Variables

Environment variables override all configuration files:

Core Variables

Variable	Description	Example
CHOOSE_CLIENT	AI provider (1=Anthropic, 2=Bedrock, 3=Gemini, 0=auto)	3
MAX_TOKEN	Maximum output tokens	4096
MAX_INTERNET_SEARCH_LIMIT	Web searches per session	5
WYN360_SKIP_CONFIRM	Skip command confirmations	1

AI Provider Credentials

Anthropic Claude

ANTHROPIC_API_KEY=your_key_here
ANTHROPIC_MODEL=claude-sonnet-4-20250514

AWS Bedrock

AWS_ACCESS_KEY_ID=your_access_key
AWS_SECRET_ACCESS_KEY=your_secret_key
AWS_SESSION_TOKEN=your_session_token
AWS_REGION=us-west-2
ANTHROPIC_MODEL=us.anthropic.claude-sonnet-4-20250514-v1:0

Google Gemini (Recommended)

GEMINI_API_KEY=your_key_here
GEMINI_MODEL=gemini-2.5-flash

OpenAI

OPENAI_API_KEY=your_key_here
OPENAI_MODEL=gpt-4o

Integration Tokens

GH_TOKEN=ghp_your_github_token              # GitHub features
GITHUB_TOKEN=ghp_your_github_token          # Alternative GitHub token var
HF_TOKEN=hf_your_huggingface_token          # HuggingFace features
HUGGINGFACE_TOKEN=hf_your_huggingface_token # Alternative HF token var

Advanced Configuration

Workspace-Specific Settings

Create different configurations for different types of projects:

# ~/.wyn360/config.yaml
# Default settings
model: "gemini-2.5-flash"
max_tokens: 4096

# Workspace-specific overrides
workspaces:
  ml_projects:
    model: "claude-sonnet-4-20250514"  # Better for complex ML tasks
    max_tokens: 8192
    custom_instructions: |
      Focus on scientific computing best practices:
      - Use NumPy/SciPy for numerical operations
      - Follow scikit-learn patterns for ML models
      - Include proper data validation and preprocessing
      - Add visualization with matplotlib/seaborn

  web_development:
    model: "gemini-2.5-flash"  # Cost-effective for web dev
    max_tokens: 4096
    browser_use_cache_enabled: true
    custom_instructions: |
      Web development focus:
      - Use modern frameworks (FastAPI, React, etc.)
      - Include proper error handling and logging
      - Follow REST API best practices
      - Include basic security measures

Aliases and Shortcuts

# ~/.wyn360/config.yaml
aliases:
  # Code review shortcuts
  review: "Review this code for bugs, performance issues, and style problems"
  security: "Analyze this code for security vulnerabilities"
  optimize: "Suggest performance optimizations for this code"

  # Documentation shortcuts
  doc: "Add comprehensive docstrings and comments to this code"
  readme: "Create a README.md file for this project"

  # Testing shortcuts
  test: "Generate comprehensive pytest tests for this code"
  integration: "Create integration tests for this API endpoint"

  # Deployment shortcuts
  deploy: "Help me deploy this application to production"
  docker: "Create a Dockerfile for this application"

Usage:

You: @review main.py
# Expands to: "Review this code for bugs, performance issues, and style problems" + file content

Configuration Commands

View Current Configuration

# In WYN360 CLI session
/config

# Shows:
# 📋 Current Configuration
# Model: gemini-2.5-flash
# Max Tokens: 4096
# Custom Instructions: [First 100 chars...]
# Project Context: [First 100 chars...]
# Browser Cache: Enabled (TTL: 1800s)

Validate Configuration

# Check config file syntax
poetry run python -c "
import yaml
with open('.wyn360.yaml') as f:
    config = yaml.safe_load(f)
    print('✅ Configuration is valid')
"

Configuration Best Practices

1. Layer Your Settings

• User config: Personal preferences, API keys, general coding style
• Project config: Project-specific context, model selection, team standards
• Environment variables: Runtime overrides, CI/CD settings

2. Use Project Context Effectively

project_context: |
  Tech Stack:
  - Backend: Django 4.2 with DRF
  - Database: PostgreSQL 14
  - Cache: Redis
  - Testing: pytest + factory_boy
  - Deployment: Docker + AWS ECS

  Project Structure:
  - /apps/core/ - Core business logic
  - /apps/api/ - REST API endpoints
  - /apps/users/ - User management
  - /config/ - Settings and configuration

  Coding Standards:
  - Follow Django best practices
  - Use class-based views for complex logic
  - Prefer function-based views for simple endpoints
  - All models must have proper __str__ methods
  - Use Django's built-in User model extensions

3. Cost Optimization

# Cost-conscious configuration
model: "gemini-2.5-flash"  # 40x cheaper than Claude
max_tokens: 2048           # Reduce for simpler tasks
browser_use_cache_enabled: true  # Avoid re-fetching pages
browser_use_cache_ttl: 3600      # Cache for 1 hour

4. Team Consistency

Create a shared .wyn360.yaml in your project repository:

# .wyn360.yaml - Team settings
model: "claude-sonnet-4-20250514"  # Consistent model for team
max_tokens: 4096
temperature: 0.1  # Consistent, less random responses

project_context: |
  Team coding standards and project setup...

custom_instructions: |
  Team-wide instructions that everyone should follow...

# Don't include API keys in project config!
# Use environment variables or user config instead

---

Next: Features Overview →