Unbridled Intelligence
Full-strength reasoning and planning from your chosen model—execution remains approval-gated and audited.
Not Just an LLM Wrapper
terminaI is a full agentic harness with real terminal execution, multi-step planning, and graceful error recovery. It doesn't just generate text—it operates your system.
- PTY Execution
Real terminal, not simulations. Full TUI support, process control, signal handling.
- Planning Loops
Multi-step reasoning, error recovery, adaptive replanning when things go wrong.
- Tool Orchestration
Shell, files, processes, web search, and custom MCP tools—all coordinated.
Built on Tomorrow's Stack
A2A for agent coordination, MCP for extensibility, offline voice for accessibility.
A2A Protocol
Agent-to-Agent communication. Control terminaI from IDEs, desktop apps, mobile, or other agents.
Learn more →MCP Integration
Model Context Protocol for infinite extensibility. Connect to GitHub, Slack, databases, and custom tools.
Learn more →Offline Voice
Whisper for STT, Piper for TTS. Download once, runs forever offline. Your voice never leaves your device.
Learn more →Powered by the World's Most Advanced Models
Support for leading models like Gemini 1.5 Pro, GPT-4o, and Llama 3. Reason over entire repositories with massive context windows.
- ✓ Compatible with OpenAI, Anthropic, and Google APIs
- ✓ Native support for local models (Ollama, LM Studio)
- ✓ Large context window handling
- ✓ Optimized for tool-calling and code generation
- ✓ Pure OSS—no model-specific lock-in