Index / QwenPaw · updated Aug 17, 2026
QwenPaw
agentscope-ai/CoPaw · healthy · rank 5 of 43 by stars
A polished, multi-chat-app personal AI assistant from the AgentScope ecosystem with strong local-first deployment and a rapidly iterating codebase. Recent commits show active hardening of memory, sandboxing, and console UX—signaling a maturing product rather than a toy prototype.
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 · 111 files examined
5 providers · runs locally · gateway support
Pinned models
claude-opus-4-5 released 2025-11-24 — 9 months old (from the public model catalogue)
Pin last edited 20 days ago
Evidence
Decision
Why choose QwenPaw over OpenClaw?
Why choose this
- Native DingTalk integration for Chinese enterprise users — OpenClaw lacks this
- AgentScope ecosystem alignment gives it a distinct agent-orchestration pedigree
- Apache-2.0 license is more permissive than some OpenClaw components
Tradeoffs
- Python runtime means higher memory usage and slower boot vs OpenClaw's Rust/Go options
- Smaller plugin ecosystem — OpenClaw has 86+ repos and a broader skill marketplace
- No evidence of edge-optimized variants (no 'pico'/'nano' tier like picoclaw)
Best fit
- Users wanting a self-hosted AI assistant that plugs into multiple chat apps (Discord, DingTalk, etc.)
- Developers who prefer Python extensibility and the AgentScope ecosystem
- Teams needing cloud-or-local deployment flexibility with sandboxed execution
Avoid if
- You need ultra-low-latency edge deployment (Python boot overhead is real)
- You want a pure Rust/Go stack with minimal memory footprint
- You require strong privacy guarantees out of the box — telemetry posture is unclear
Evidence is moderate: README and commit history are rich and recent, confirming active development, but Reddit results are entirely noise (pet/cat posts matching 'paw'), and web search results are thin beyond GitHub listings. Community sentiment cannot be reliably scored from the provided data.
AI layer reviewed Aug 2, 2026 · how this is written
Star activity
33,785 stars today
Overview
QwenPaw is a personal AI assistant framework developed under the AgentScope-AI organization, designed for easy installation and deployment on local machines or in the cloud. It supports multiple chat applications — including Discord and DingTalk — and emphasizes extensibility through a Python-based plugin architecture. The project is licensed under Apache-2.0 and targets Python 3.11 through 3.14, with a codebase that enforces Black formatting and maintains active CI/CD via PyPI releases.
Architecturally, QwenPaw appears to follow a modular agent design pattern common in the AgentScope ecosystem. Recent commits reveal a sophisticated memory subsystem (Auto-Memory with Scroll context eviction and a ReMe knowledge base), sandbox mode with auto-downgrade safety, ACP protocol support for session management, and git identity injection for coding workflows. The v2.0.1 release and the rapid cadence of fixes (audio transcription, console session integrity, skill injection) suggest a project in active hardening — moving from feature velocity toward production reliability.
Compared to OpenClaw, QwenPaw differentiates through its AgentScope lineage and Chinese-platform integrations (DingTalk), but shares the same core vision of a self-hosted, multi-channel AI assistant. Where OpenClaw has diversified into Rust-based edge variants (picoclaw) and safety guardrails (NemoClaw), QwenPaw remains a Python-first monolith with an emerging plugin ecosystem. Its sandboxing and memory work are notable strengths, but the Python runtime imposes inherent latency and memory tradeoffs versus OpenClaw's compiled-language options.