Strategic AI Agent for the Terminal
A terminal-native autonomous AI agent built on a Strategic Agentic Loop. Connects to 7 AI providers, executes tasks through 67+ built-in tools, and streams results in real time — from file operations and live web search to OCR, PDF editing, and full browser automation.
Get Started
On first launch, set your API key with /key or open the settings panel via Ctrl+P.
Capabilities
LLM-based task decomposition with step prioritization, progress tracking, and multi-step reasoning before execution.
Category-based filtering and relevance ranking across 14 tool categories including PDF, OCR, browser automation, and MCP real-time data.
7 AI backends with seamless switching — OpenRouter, Gemini, Anthropic, OpenAI, Groq, Together, and HuggingFace.
Max 12 rounds, same-tool detection at 3x, and anti-stuck content analysis to prevent runaway execution cycles.
16 MCP tools for weather, stocks, crypto, news, prayer times, and more — all free with no keys required.
Optimized for Android Termux with raw-mode input, paste detection, and minimal resource usage.
AI Backends
| Provider | Free Tier | Best For |
|---|---|---|
| OpenRouter | FREE | DeepSeek-R1, Llama, Claude, GPT-4o — 200+ models in one endpoint |
| Google Gemini | FREE | High-speed reasoning with large context windows |
| HuggingFace | FREE | Open-source model ecosystem |
| Groq | FREE | Ultra-low latency inference |
| Together AI | FREE | Diverse open-source model selection |
| Anthropic | PAID | Precision coding and reasoning with Claude 4 |
| OpenAI | PAID | Industry-standard GPT-4o and o3 models |
REPL Interface
Tool Ecosystem
File
Web
Code
System
DOCX
OCR
Browser
MCP Real-Time
Math
Image
Utility
Internals