Olly – Your Autonomous AI Coding Assistant 🤖

Olly is a powerful autonomous AI agent built to help you with code generation, debugging, task planning, and more. Access it interactively from your terminal with 15+ AI providers including the new TokenLB provider.

✨ Features

• 15+ AI Providers — TokenLB, OpenRouter, OpenAI, Anthropic, Google Gemini, Groq, Mistral, xAI (Grok), DeepSeek, Cohere, Perplexity, Cerebras, Azure OpenAI, and Ollama (local/offline)
• 100+ Models — Pick the exact model you want, with context window info
• TokenLB Support — OpenAI-compatible proxy supporting claude-opus-4-6, GPT-4o, and more
• ⚡ Slash Commands — Type / in any interactive mode for live config changes
• No-Confirm Agent Mode — Agent mode executes tasks immediately, no "proceed?" prompts
• Free Tier Support — OpenRouter's free routing, Groq, Cerebras, and Google free tiers
• Interactive CLI Setup — Guided wizard or one-liner olly set commands
• Multiple Interfaces — Terminal CLI or Telegram Bot
• Smart Code Generation — Build features, find bugs, plan tasks, and explain code

📦 Installation

Install Olly globally via npm:

npm install -g my-olly

Bun is installed automatically if not already present. You may need to restart your terminal afterwards on Windows.

🛠️ Setup

After installation, run the setup wizard:

olly setup

Or use quick one-liner commands:

olly set provider tokenlb
olly set model claude-opus-4-6
olly set apikey sk-hUEBQXvY5zgRhgT3GYBgsYnZbEfRnLGBrAw3ImHrv0GdFelT

⚡ Slash Commands (in any mode)

When using olly agent, olly ask, or olly plan, type / to see available commands:

| Command | Description | |---|---| | /model | Switch AI model interactively | | /model gpt-4o | Switch to a specific model instantly | | /provider | Switch AI provider interactively | | /provider tokenlb | Switch to TokenLB instantly | | /config | View and edit full configuration | | /status | Show current provider, model & auto mode | | /auto | Toggle auto-approve mode | | /auto on | Enable auto-approve (no confirmations) | | /auto off | Disable auto-approve | | /approve | Set approval mode (always/ask) | | /help | List all slash commands |

💻 Usage

# Wake up Olly (main interface)
olly wakeup

# Agent mode (executes immediately, no confirmations)
olly agent
olly agent "Refactor index.ts to use async/await"

# Ask mode (read-only Q&A)
olly ask
olly ask "What packages are in my project?"

# Plan mode (generate & execute step-by-step plans)
olly plan

# Quick config without re-running setup
olly set provider tokenlb
olly set model claude-opus-4-6
olly set auto on
olly config       # interactive config editor
olly status       # show current configuration

# Other commands
olly setup        # full interactive setup wizard
olly doctor       # check config and AI connection
olly version      # show version and current config

🔌 Supported Providers

| Provider | Key env var | Notes | |---|---|---| | TokenLB | TOKENLB_API_KEY | OpenAI-compatible, Claude Opus 4.6 | | OpenRouter | OPENROUTER_API_KEY | ✅ Free tier (openrouter/free) | | OpenAI | OPENAI_API_KEY | GPT-4.1, o3, o4-mini | | Anthropic | ANTHROPIC_API_KEY | Claude Opus/Sonnet/Haiku | | Google Gemini | GOOGLE_GENERATIVE_AI_API_KEY | ✅ Free tier | | Groq | GROQ_API_KEY | ✅ Free tier, ultra-fast | | Mistral | MISTRAL_API_KEY | Mistral Large, Medium | | xAI (Grok) | XAI_API_KEY | Grok-4, Grok-3 | | DeepSeek | DEEPSEEK_API_KEY | DeepSeek V3, R1 | | Cohere | COHERE_API_KEY | ✅ Trial credits | | Perplexity | PERPLEXITY_API_KEY | Sonar models | | Cerebras | CEREBRAS_API_KEY | ✅ Free tier, ultra-fast | | Azure OpenAI | AZURE_API_KEY | Your own deployment | | Ollama | (no key, local) | ✅ 100% free, offline |

📁 Configuration

All keys are stored in ~/.olly/.env. Core variables:

OLLY_PROVIDER='tokenlb'                # provider id
OLLY_MODEL='claude-opus-4-6'           # model id
TOKENLB_API_KEY='sk-hU...'            # provider API key
OLLY_AUTO='1'                         # auto-approve all (optional)

🔗 Links

• OpenRouter — aggregator with free tier
• Groq Console — fastest free inference
• Google AI Studio — free Gemini API keys
• Cerebras — ultra-fast free inference
• Ollama — run models locally for free
• TokenLB — OpenAI-compatible proxy