Core Coder V5

A custom CLI agent

Overview

This is a custom CLI agent created with Agent Forge. It provides automated assistance and can be extended with custom tools and knowledge.

Installation

1. Ensure you have the langchain-agent-base framework available:

# Clone or ensure langchain-agent-base is in parent directory
cd path/to/langchain-agent-base

# Activate virtual environment
.\\venv\\Scripts\\Activate.ps1  # Windows PowerShell
# or
source venv/bin/activate        # Linux/Mac

# Install in development mode
pip install -e .

2. Set up your environment variables:

cp .env.template .env
# Edit .env and add your GROQ_API_KEY

3. Place this agent directory alongside langchain-agent-base

Usage

Interactive Mode (Recommended)

python cli.py

With a Specific Project Directory

python cli.py /path/to/project

Without Memory

python cli.py --no-memory

CLI Commands

All //commands are implemented using the CommandRegistry system from langchain-agent-base.

While in interactive mode:

• //help - Show all available commands
• //tools - List all available tools
• //status - Show agent status
• //config - Show full configuration
• //model - Hot-swap provider and model (no restart needed!)
• //memory status|clear|show - Manage conversation memory
• //rag status|search - RAG knowledge base
• //ollama list - List available Ollama models
• //groq list - List available Groq models

Custom commands can be added by editing langchain-agent-base/src/commands.py.

Knowledge Base

This agent includes 3 knowledge files across 4 categories:

• constitution: Domain-specific knowledge for constitution
• technical: Domain-specific knowledge for technical
• design: Domain-specific knowledge for design
• custom: Domain-specific knowledge for custom

Architecture

This agent uses a template-based architecture for maximum maintainability:

• cli.py: Copied from langchain-agent-base/src/cli_template.py
• agent_config.json: All agent-specific configuration (single source of truth)
• //commands: Defined in langchain-agent-base/src/commands.py using CommandRegistry
• langchain-agent-base: Core framework with tools, memory, RAG, and middleware

This design means:

• ✅ No code duplication between JS and Python
• ✅ Updates to langchain-agent-base automatically improve all agents
• ✅ Easy to test and maintain commands in one place
• ✅ All customization via JSON configuration

Framework Features

• ⚡ Groq Integration: 10x faster inference with gpt-oss-120b
• 🧠 Persistent Memory: Conversation history with Qdrant storage
• 📚 RAG Capabilities: Document search and knowledge integration
• 🔧 Comprehensive Tools: File operations, web search, calculations
• 🛡️ Security: Built-in safety controls and middleware
• 🔄 Multi-Agent: Can coordinate with other specialized agents

Configuration

All settings are stored in agent_config.json. You can edit it directly or use CLI commands.

The config includes:

• Model settings (provider, model_name, temperature)
• Feature flags (enable_shell, enable_memory, enable_rag)
• Toolbox selections
• CLI appearance and ASCII art
• System prompt

Customization

To customize the agent:

1. Edit Configuration: Modify agent_config.json directly
2. Add Tools: Create new functions with the @tool decorator
3. Extend Capabilities: Add more tools to the tools list
4. Add Knowledge: Upload documents to the knowledge/ directory

Support

For issues or questions:

1. Check the langchain-agent-base documentation
2. Review the examples
3. Test with the comprehensive test suite

---

Generated by Agent Forge - Visual Agent Builder