Index / LightClaw · updated Oct 2, 2026
LightClaw
zofrasca/lightclaw · healthy · rank 36 of 43 by stars
A single-binary Rust rewrite of the OpenClaw concept, shipping agentic tooling, memory, and Telegram/Discord integration in a ~15MB executable. It positions itself as the fast, local-first, low-footprint alternative for users who want Claw without the heavy runtime.
Facts
Repository
Runtime
Posture
Community
Security breakdown
Composite 62 / 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 · 35 files examined
2 providers · runs locally · gateway support
Pinned models
claude-opus-4-5 released 2025-11-24 — 10 months old (from the public model catalogue)
Pin last edited 8 months ago
Evidence
Decision
Why choose LightClaw over OpenClaw?
Why choose this
- ~15MB single binary vs heavy multi-component install
- Near-instant startup and low memory footprint
- Broad platform support including ARMv7 and Raspberry Pi
Tradeoffs
- Smaller community and ecosystem than OpenClaw
- Windows support is unstable and second-class
- Fewer integrations and no full control UI
Best fit
- Users wanting a single-binary, low-footprint AI assistant
- Raspberry Pi / ARM and low-resource device deployments
- Telegram/Discord-native personal assistant setups
Avoid if
- You need a mature plugin ecosystem or web dashboard
- You require Windows as a first-class platform
- You want battle-tested sandboxing around shell tools
Evidence is solid on architecture and positioning from the README and recent commits, but community sentiment data is thin (one irrelevant Reddit match, mostly ecosystem-level web results), so popularity and sentiment scores carry real uncertainty.
AI layer reviewed Sep 21, 2026 · how this is written
Star activity
223 stars today
Overview
LightClaw is a Rust-based, single-binary reimagining of OpenClaw, inspired by both OpenClaw and HKUDS's nanobot. It packages an agentic assistant — with file, shell, web, and scheduling tools — into a ~15MB executable with no Python runtime or heavy dependencies, targeting near-instant startup and a minimal memory footprint. It is powered by the Rig framework and supports OpenClaw-style SKILL.md skills via an activate_skill mechanism.
The project ships a local-first memory system built on SQLite: per-session short-term history, periodic summarization of conversation chunks, and semantic retrieval over stored memories, all without an external vector database. It integrates natively with Telegram and Discord via high-performance polling, supports multiple LLM providers (OpenRouter, OpenAI, Ollama, Mistral) with model fallbacks, and recently added XDG base directory support and ARMv7 builds for devices like older ARM boards and e-readers.
Compared to OpenClaw, LightClaw trades ecosystem breadth and UI surface for deployability and footprint: one binary, one config file, and broad platform coverage (Linux x86_64/ARM64/ARMv7, macOS Intel/Apple Silicon, with Windows present but less stable). Recent commits show active maintenance — rebranding from femtobot, shell-guard fixes, and firecrawl configuration — suggesting a young but moving project best suited to self-hosters who value simplicity over feature depth.