Skip to content
BranchesTags

Repository files navigation

MyLocalCLI ⚡

Your Own AI Coding Assistant - Private, Local, Yours

A Claude Code alternative that works with local LLMs and free cloud APIs. Now with agents, skills, project config, and cross-platform support!

npm Node.js License Platform

✨ Features

Feature Description
🏠 6 AI Providers LM Studio, Ollama, OpenRouter, OpenAI, Groq, Custom
🛠️ 26 Tools File ops, search, git, web fetch, todos, multi-edit
🤖 5 Agents Code reviewer, explorer, test generator, refactorer, doc writer
🎓 22 Skills Auto-injected best practices for JS, Python, React, and more
📋 15+ Commands Slash commands like Claude Code
📝 Project Config MYLOCALCLI.md for project-specific instructions
🌐 Web UI Beautiful dark theme with voice input
🔄 Cross-Platform Works on Windows, macOS, and Linux
🔒 Private Runs locally, your data stays yours

🚀 Installation

npm install -g mylocalcli

Now run (both commands work):

mlc init          # or: mylocalcli init
mlc               # or: mylocalcli

🎯 Quick Start

1. Setup (First Time)

mlc init

This wizard helps you:

  • Choose an AI provider (LM Studio, Ollama, OpenRouter, etc.)
  • Configure API endpoints and keys
  • Select a model

2. Start Chatting

mlc

3. Try the Web UI

mlc web

Open http://localhost:3000 in your browser.

📖 Usage Guide

Basic Chat

Just type your question or request:

You: Explain this function
You: Fix the bug in src/utils.js
You: Create a REST API for user authentication

Slash Commands

/help         - Show all commands
/tools        - List 26 available tools
/agents       - List 5 specialized agents  
/skills       - List 22 auto-injecting skills
/init-config  - Create MYLOCALCLI.md project config
/provider     - Switch AI provider
/model        - Switch model
/models       - List available models
/history      - View saved conversations
/clear        - Clear conversation
/exit         - Exit the chat

Multi-line Input

Start with triple backticks for code blocks:

You: ```
function add(a, b) {
  return a + b;
}
```
Explain this code

🛠️ Tools (26)

MyLocalCLI has 26 built-in tools the AI can use:

File Operations

Tool Description
read_file Read file contents
write_file Create or overwrite files
edit_file Edit specific parts of files (fuzzy matching)
multi_edit_file Make multiple edits in one operation
delete_file Delete files or directories
copy_file Copy files
move_file Move or rename files
file_info Get file metadata (size, dates)
append_file Append content to files
insert_at_line Insert at specific line number
read_lines Read specific line range

Search & Navigation

Tool Description
list_directory List directory contents
search_files Find files by glob pattern
grep Search text in files
tree Show directory structure
find_replace Find and replace across files
codebase_search Semantic code search

Git Operations

Tool Description
git_status Get repository status
git_diff Show changes
git_log Show commit history
git_commit Create commits

Other

Tool Description
run_command Execute shell commands (cross-platform!)
web_fetch Fetch content from URLs
todo_write Maintain task lists
ask_user Ask user for input/confirmation
create_directory Create directories

Cross-Platform Commands

Commands like ls, cat, rm are automatically translated on Windows:

Unix → Windows
ls   → dir
cat  → type
rm   → del
cp   → copy
mv   → move
pwd  → cd

🤖 Agents (5)

Agents are specialized personas for specific tasks:

Agent Description Example
code-reviewer Reviews code for bugs, security, style /agent code-reviewer Review auth.js
code-explorer Deep codebase analysis /agent code-explorer How does the auth flow work?
test-generator Generates unit tests /agent test-generator Create tests for utils.js
refactorer Suggests improvements /agent refactorer Refactor the User class
doc-writer Generates documentation /agent doc-writer Document the API endpoints

🎓 Skills (22)

Skills automatically inject best practices based on your project files:

Languages

Skill Triggers On Priority
JavaScript *.js, *.ts, *.jsx, *.tsx 100
Python *.py, pyproject.toml 100
Rust *.rs, Cargo.toml 90
Go *.go, go.mod 90

Frameworks

Skill Triggers On Priority
React *.jsx, *.tsx 95
Vue *.vue 95
Next.js next.config.*, app/** 90
Express server.js, routes/** 85
Django settings.py, views.py 90
FastAPI main.py, routers/** 90

DevOps & Databases

Skill Triggers On Priority
Docker Dockerfile, docker-compose.yml 80
Kubernetes k8s/**/*.yaml 70
CI/CD .github/workflows/*.yml 75
SQL *.sql, migrations/** 80
MongoDB models/**/*.js 75
Redis redis*.js, cache*.js 70

Best Practices

Skill Triggers On Priority
Security All code files, SECURITY.md 100
Testing *.test.js, *.spec.ts 85
Git Workflow .git/**, CONTRIBUTING.md 80
API Design routes/**, api/** 85
Performance *.html, *.css, *.js 75
Node.js package.json, index.js 90

Custom Skills

Create your own skills in .mylocalcli/skills/<name>/SKILL.md:

mlc
> /init-skill my-framework

Or manually:

---
name: my-framework
description: Best practices for My Framework
globs: ["**/*.myf"]
priority: 50
tags: ["custom"]
---

# My Framework Best Practices

- Guideline 1
- Guideline 2

📝 Project Configuration

Create a MYLOCALCLI.md file in your project root to give the AI project-specific instructions:

mlc
> /init-config

Example MYLOCALCLI.md:

---
name: My Project
description: A Node.js API server
author: Your Name
---

# Project Instructions

- Use TypeScript for all new files
- Follow REST API conventions  
- Write tests for all endpoints
- Use Prisma for database access
- Follow conventional commits

# Coding Standards

- Use ESLint and Prettier
- Maximum function length: 50 lines
- Always handle errors with try/catch

# File Structure

src/ ├── routes/ # API routes ├── services/ # Business logic ├── models/ # Database models └── utils/ # Helpers

🤖 Supported Providers

Provider Type Free? Setup
🏠 LM Studio Local ✅ Download LM Studio → Load model → Start server
🦙 Ollama Local ✅ ollama pull llama3.2 && ollama serve
🌐 OpenRouter Cloud ✅ Get free API key from openrouter.ai
⚡ Groq Cloud ✅ Get free API key from console.groq.com
🔑 OpenAI Cloud ❌ Requires paid API key
⚙️ Custom Any - Any OpenAI-compatible endpoint

Recommended Free Setup

Option 1: Local (Privacy) - LM Studio

# 1. Download LM Studio from https://lmstudio.ai
# 2. Load a model (e.g., Qwen 2.5 Coder 7B)
# 3. Start Local Server (port 1234)
mlc init  # Select LM Studio

Option 2: Local (Lightweight) - Ollama

# Install Ollama from https://ollama.ai
ollama pull llama3.2
ollama serve
mlc init  # Select Ollama

Option 3: Cloud (Free) - OpenRouter

# Get free API key from https://openrouter.ai
mlc init  # Select OpenRouter → Enter API key

🌐 Web UI

mlc web
# Opens http://localhost:3000

Features:

  • 🌙 Beautiful dark theme
  • 🎤 Voice input support
  • 💬 Conversation history
  • 🔄 Provider/model switching
  • 📱 Mobile-friendly

⌨️ CLI Features

Feature Description
Input History Press ↑/↓ to navigate previous commands
Tab Completion Type / then Tab for command suggestions
Multi-line Input Start with ``` for code blocks
Streaming Real-time response display
Token Counter See context usage percentage
Auto-approval Use mlc --auto for unattended operation

🔧 Configuration

Configuration is stored in ~/.mylocalcli/:

~/.mylocalcli/
├── config.json      # Provider settings
├── history/         # Conversation history
└── skills/          # Custom skills (global)

Project-local configuration:

your-project/
├── MYLOCALCLI.md       # Project instructions
└── .mylocalcli/
    └── skills/         # Project-specific skills

🛡️ Privacy & Security

⚠️ Important Privacy Notice

Local Providers (Full Privacy ✅)

  • LM Studio, Ollama: All data stays on YOUR machine
  • No data leaves your computer - 100% private
  • Recommended for sensitive code

Cloud Providers (Data Shared ⚠️)

  • OpenRouter, Groq, OpenAI: Your code/prompts ARE sent to their servers
  • These providers may log or store your data per their privacy policies
  • Good for non-sensitive projects or trying the tool quickly

For Maximum Privacy

If you need full privacy with cloud-grade performance, run your own model:

# Option 1: vLLM (GPU required)
pip install vllm
vllm serve meta-llama/Llama-3.1-8B-Instruct --port 8000

# Option 2: Use vLLM with Custom provider in mlc
mlc init  # Select "Custom" → Enter http://localhost:8000/v1

Quick Start Links

Provider Privacy Link
🏠 LM Studio ✅ Full lmstudio.ai
🦙 Ollama ✅ Full ollama.ai
⚡ vLLM (Self-hosted) ✅ Full docs.vllm.ai
🌐 OpenRouter ⚠️ Cloud openrouter.ai
⚡ Groq ⚠️ Cloud console.groq.com

Security Best Practices

  • API keys: Stored locally in ~/.mylocalcli/config.json
  • No telemetry: MyLocalCLI does NOT collect any usage data
  • Open source: Audit the code yourself on GitHub

🐛 Troubleshooting

"Connection refused" error

  • Make sure LM Studio/Ollama server is running
  • Check the port (default: 1234 for LM Studio, 11434 for Ollama)

Command not found on Windows

  • Unix commands are auto-translated (ls → dir)
  • If it still fails, use Windows commands directly

Slow responses

  • Try a smaller model
  • Use Groq for fast cloud inference
  • Reduce context with /clear

Model not loading

  • Check available disk space
  • Verify model compatibility with your hardware

🙏 Credits

Built with the assistance of:

📄 License

MIT - Use it, modify it, make it yours!

Made with ❤️ by Prashanth Kumar

⭐ Star this repo if you find it useful!

About

No description, website, or topics provided.

Resources

Readme
MIT license

Contributing

Contributing
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

Packages

Contributors

Languages