Skip to main content

Introduction

SuprClaw is an ultra-lightweight personal AI assistant written in Go.

Runs on $10 hardware with <10MB RAM. Single binary, 1-second boot, works across x86_64, ARM64, MIPS, and RISC-V.

Why SuprClaw?​

Most AI assistants require Electron, Node.js, or Docker with hundreds of megabytes of dependencies. SuprClaw is a single Go binary that:

  • Boots in 1 second on a 0.6GHz single-core board
  • Uses less than 10MB RAM — 99% lighter than Electron alternatives
  • Has no runtime dependencies — drop it anywhere and run
  • Works on any architecture — x86_64, ARM64, ARMv7, MIPS, RISC-V, LoongArch
  • Keeps your data local — no telemetry, no tracking

Key Features​

FeatureDescription
Multi-ChannelTelegram, Discord, WhatsApp, Matrix, LINE, and more
Multi-ProviderAnthropic, OpenAI, Gemini, Groq, Ollama, OpenRouter, Azure, and more
Scheduled TasksBuilt-in cron scheduler
MCP SupportModel Context Protocol integration with lazy tool loading
Skills SystemExtend agent capabilities with custom skills
Security SandboxAgent restricted to workspace by default
Credential EncryptionAES-256-GCM encrypted API keys in config
Web SearchPerplexity, Brave, SearXNG, DuckDuckGo
Voice TranscriptionAudio messages auto-transcribed via Groq Whisper

How It Works​

You ──► Chat Channel ──► Gateway ──► Agent ──► AI Provider
         (Telegram,         │          │         (Anthropic,
          Discord,          │          │          OpenAI, ...)
          WhatsApp, ...)     │          │
                            │          └──► Tools (web search,
                            │               file ops, exec, ...)
                            └──► Web Console (optional)

The gateway receives messages from all configured chat channels, routes them to an agent, which calls the configured AI provider and available tools, then sends the response back.

Next Steps​