Index / SmallClaw · updated Oct 24, 2026
SmallClaw
XposeMarket/SmallClaw · healthy · rank 41 of 50 by stars
A local-first, chat-first AI agent framework tuned for small local models, with a single-pass architecture that skips multi-agent planning overhead. It pairs a clean web UI with real tools (files, browser, terminal, web search) and optional hybrid cloud providers.
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 · 17 files examined
1 provider · runs locally
Pinned models
gpt-4o released 2024-05-13 — 2 years old (from the public model catalogue)
Pin last edited 8 months ago
Several defaults are pinned; the project picks one per provider rather than shipping a single default.
Evidence
Decision
Why choose SmallClaw over OpenClaw?
Why choose this
- Single-pass chat handler reduces latency vs multi-agent pipelines
- Designed specifically for small local models that struggle with complex orchestration
- No API costs with fully local provider support
Tradeoffs
- Far smaller community and ecosystem than OpenClaw
- No multi-channel messaging breadth (WhatsApp, Telegram, Slack, etc.)
- No official releases published on GitHub
Best fit
- Users running small local models via Ollama or llama.cpp
- Privacy-conscious users wanting a chat-first local agent UI
- Tinkerers who want drop-in SKILL.md extensibility
Avoid if
- You need multi-channel messaging (WhatsApp, Slack, etc.) at OpenClaw scale
- You require a mature plugin ecosystem and large community
- You want a proven security track record and formal releases
Evidence is moderate: the README is detailed and commits are recent, but there are no official releases, no genuine Reddit discussion (matches are unrelated noise), and the license is ambiguous. Sentiment and adoption signals are weak.
AI layer reviewed Oct 5, 2026 · how this is written
Star activity
259 stars today
Overview
SmallClaw is a local-first AI agent framework from Xpose Market, positioned as an open-source alternative to cloud AI assistants. It is chat-first: a single-pass LLM call decides whether to respond conversationally or invoke tools, avoiding the separate planning/execution/verification agent loops that small models handle poorly. This design choice directly targets users running Ollama, llama.cpp, or LM Studio on consumer hardware, with optional hybrid support for OpenAI API and Codex OAuth.
Feature-wise it is surprisingly complete for its size: line-precise file editing, multi-provider web search (Tavily, Google, Brave, DuckDuckGo), Playwright-powered browser automation, terminal access, persistent session memory with pinned context, and a drop-in SKILL.md skills system. Recent commits show active development through v1.1, including security hardening, Docker improvements, background tasks, sub-agents, a Telegram file browser, MCP/webhook support, and a task system with dark mode UI.
Compared to OpenClaw, SmallClaw trades ecosystem breadth for simplicity and latency. It lacks OpenClaw's multi-channel messaging surface and massive skills registry, but its single-pass architecture is arguably better suited to weak local models. Caveats: no official GitHub releases, an ambiguous license ('All rights reserved' in the description despite an open-source claim), and essentially no genuine community discussion found in search results.