Index / zclaw · updated Sep 4, 2026
zclaw
tnm/zclaw · healthy · rank 21 of 43 by stars
An ultra-minimalist AI personal assistant that squeezes onto an ESP32 microcontroller with an all-in firmware budget of 888 KiB. It brings GPIO control, cron scheduling, persistent memory, and custom tools to bare-metal hardware.
Facts
Repository
Runtime
Posture
Community
Security breakdown
Composite 72 / 100 · how these are scored
higher is safer
higher is safer
higher is safer
higher is safer
higher is riskier
Model access
Read from the repository, not written by a model · 57 files examined
2 providers · runs locally · gateway support
Pinned models
claude-sonnet-4-6 released 2026-02-17 — 7 months old (from the public model catalogue)
Pin last edited 6 months ago
Several defaults are pinned; the project picks one per provider rather than shipping a single default.
Evidence
Decision
Why choose zclaw over OpenClaw?
Why choose this
- Runs on $5 ESP32 hardware with 888 KiB total firmware
- Direct GPIO, I2C, and DHT sensor control built in
- Near-zero power draw for always-on deployments
Tradeoffs
- Requires external LLM API or relay for intelligence
- Far smaller tool and skill ecosystem
- Single-user, single-device scope
Best fit
- Makers wanting an AI assistant on ESP32 hardware
- GPIO and sensor automation via natural language
- Ultra-low-power, always-on edge deployments
Avoid if
- You need a full desktop or server agent
- You want rich plugin ecosystems or multi-user support
- You lack ESP32 hardware or flashing experience
Strong evidence from README, docs, and recent commits confirms an active, well-documented embedded assistant. Community sentiment is uncertain due to zero Reddit matches and only indirect web mentions.
AI layer reviewed Aug 24, 2026 · how this is written
Star activity
2,223 stars today
Overview
zclaw is a C-based AI personal assistant designed to run entirely on ESP32 microcontrollers, with a strict all-in firmware budget of 888 KiB that includes the ESP-IDF/FreeRTOS runtime, Wi-Fi, TLS, and certificate bundles. Despite its tiny footprint, it supports timezone-aware scheduled tasks (daily, periodic, one-shot), GPIO control, persistent memory, and user-defined tools composed through natural language. Interaction happens via Telegram or a hosted web relay, with LLM backends configurable at provisioning time (including Ollama for local inference).
The project emphasizes hackability and security-conscious defaults: encrypted credential storage in flash via secure flash mode, compile-time LLM rate limits, Telegram chat ID allowlists, and bootstrap integrity checks. Recent releases added hardware-oriented tools like DHT sensors and generic I2C, reinforcing its positioning as a bridge between conversational AI and physical computing.
Compared to OpenClaw, zclaw trades ecosystem breadth and desktop-class capability for extreme minimalism and hardware intimacy. It is not a general-purpose agent platform but rather the smallest viable embodiment of the personal-assistant concept, ideal for makers who want an always-on AI companion wired directly into sensors, relays, and the physical world.