Forge Code Integration

Overview

Forge Code is an AI-powered shell assistant that brings intelligent coding capabilities directly to your terminal. Unlike IDE-specific tools, Forge works seamlessly with any development environment, making it the perfect companion for developers who value their existing workflow.

Key Features

Terminal-Native: Works in your shell without requiring a specific IDE
Model Agnostic: Supports 300+ models including RelaxAI, OpenAI, Anthropic Claude, and local models
Privacy-First: All code stays on your machine with your own API keys
Zero IDE Lock-in: Compatible with VS Code, Neovim, IntelliJ, Xcode, or any editor
Git-Aware: Understands your entire codebase and version history
Custom Workflows: Create and share specialized agents for different tasks

Use Cases

Understanding complex codebases and legacy code
Implementing new features with natural language descriptions
Debugging issues and finding root causes
Performing large-scale refactors and migrations
Generating tests and documentation
Answering questions about code architecture and data flow

Use Case

Perfect for developers who prefer working in the terminal and want AI assistance without switching contexts or learning new IDEs.

Setup Instructions

Install Forge globally
Terminal window
```
npm install -g forgecode@latest
```
Create environment configuration

Create a .env file in your home directory with the following configuration:
Terminal window
```
# API Configuration for RelaxAI
OPENAI_API_KEY=RELAX_API_KEY
OPENAI_URL=https://api.relax.ai/v1
```
Configure Forge to use RelaxAI
Terminal window
```
npx forgecode@latest
```

Select your model

Select the model from the list of available models. You can use RelaxAI models like Llama-4-Maverick-17B-128E or any other compatible model.

❯ /model
? Select a model:
DeepSeek-R1-0528
Mistral-7b-embedding
➤ Llama-4-Maverick-17B-128E
DSE-QWen2-2b-MRL-V1
[Type a name or use arrow keys to navigate and Enter to select]

Start using Forge

Start using Forge in your terminal by typing in any natural language command, or experiment with the interactive agents. Please refer to the Forge Commands for more details on available commands and configurations.
Terminal window
```
# forge shell
> "Implement a new feature that allows users to export data in CSV format"
```
Create custom workflows (optional)

You can also create custom workflows by defining agents in the forge.yaml file in your project root. This allows you to tailor Forge’s behavior to your specific development needs. An example command to use the muse agent for planning and implementing a feature:

Commands and Usage

Forge provides both built-in commands and natural language interaction:

Built-in Commands:

/new - Start a new task with fresh context
/info - View environment summary and configuration details
/model - Select and set a specific AI model
/agent - Switch between available agents
/forge - Switch to Forge Agent (full execution capabilities)
/muse - Switch to Muse Agent (read-only analysis mode)
/dump - Save current conversation in JSON format

Shell Commands: Execute native shell commands directly by prefixing with !:

!ls -la - List files
!python script.py - Run scripts

Natural Language Interaction: Simply type your requests in natural language:

“Explain this function and how it works”
“Debug this error message…”
“Refactor this code to improve…”

Advanced Features:

Agent Modes: Switch between Forge (execution) and Muse (analysis) agents Context Understanding: Automatically analyzes your project structure Git Integration: Understands code history and changes Multi-file Operations: Handle complex changes across multiple files

Best Practices

Keep your prompts clear and specific for best results
Use Forge for complex tasks that benefit from codebase understanding
Review generated code before committing
Create custom workflows for repetitive tasks
Leverage git integration for understanding code evolution

Troubleshooting

Connection Issues: Verify your API endpoint and key in the .env file
Slow Responses: Consider using a faster model or adjusting max_tokens
Context Errors: Try reducing the scope of your request or breaking it into smaller tasks