{"slug":"qwenpaw","name":"QwenPaw","vendor":"AgentScope (Alibaba)","tagline":"AgentScope's self-hosted personal AI assistant with three-layer memory, kernel-level sandboxing and every chat channel","url":"https://agentboards.org/agent/qwenpaw","website":"https://qwenpaw.agentscope.io/","docs":"https://qwenpaw.agentscope.io/","repo":"https://github.com/agentscope-ai/QwenPaw","category":"harness","execution":["local","cloud"],"license":"Apache-2.0","open_source":true,"pricing":{"model":"byok","summary":"Free and Apache-2.0 self-hosted with your own model keys; optional cloud deployment on the AgentScope Platform or ModelScope Studio","from_usd":0,"free_tier":true,"byok":true},"models":{"backbone":["Qwen","OpenAI","Anthropic","Gemini","DeepSeek","Kimi","OpenRouter","QwenPaw-Flash","Ollama","LM Studio"],"bring_your_own_model":true,"local_models":true},"protocols":{"mcp_client":true,"mcp_server":false,"openapi":true},"capabilities":{"terminal_exec":true,"browser":true,"multi_file_edit":false,"git_ops":false,"docker_sandbox":true,"multi_agent":true,"headless_ci":false},"context_window":null,"platforms":["macos","linux","windows"],"install":[{"label":"pip","command":"pip install qwenpaw && qwenpaw init --defaults && qwenpaw app"},{"label":"curl (macOS, Linux)","command":"curl -fsSL https://qwenpaw.agentscope.io/install.sh | bash"},{"label":"docker","command":"docker run -p 127.0.0.1:8088:8088 -v qwenpaw-data:/app/working agentscope/qwenpaw:latest"}],"benchmarks":[],"tags":["open-source","personal-agent","memory","sandbox","messaging","mcp","acp","alibaba","agentscope"],"github_stars":35413,"metrics":{"stars":35413,"pushed_at":"2026-09-30T08:35:50Z","archived":false,"open_issues":1009,"latest_version":"v2.2.1","latest_version_at":"2026-09-11T04:19:23Z","latest_version_url":"https://github.com/agentscope-ai/QwenPaw/releases/tag/v2.2.1","version_source":"github","checked_at":"2026-10-02T05:02:20.970Z"},"sources":{"readme":"https://github.com/agentscope-ai/QwenPaw","install":"https://github.com/agentscope-ai/QwenPaw","capabilities":"https://github.com/agentscope-ai/QwenPaw"},"verified_at":null,"adoption":{"score":6.9,"signals":{"pypi_weekly":2276,"github_stars":35284},"measured_at":"2026-09-28"},"scores":{"panel":6.8,"adoption":6.9,"spread":2,"facts":2.3,"board":6.1,"dimensions":{"reliability":6.3,"usefulness":6.7,"cost":7.3,"longevity":6.7},"per_persona":{"amigo":7.3,"critico":6,"profesor":6.5,"inversora":6.8,"jefa":6,"hacker":8},"community":null},"rank":null,"category_rank":60,"panel_reviews":[{"persona":"juez","persona_name":"El Juez","ai_generated":true,"verdict":"Two points across the panel and no dispute about the sandboxing; the split is El Hacker's Tool Guard against El Crítico's fifteen roadmap items still in progress.","scores":null,"evidence":[]},{"persona":"amigo","persona_name":"El Amigo","ai_generated":true,"verdict":"Pick QwenPaw if you want a personal agent in DingTalk, Lark or iMessage that runs on its own small models without a key; pick NanoClaw if you would rather have Docker walls and Claude.","scores":{"reliability":6,"usefulness":7,"cost":9,"longevity":7},"evidence":["https://github.com/agentscope-ai/QwenPaw","https://qwenpaw.agentscope.io/"]},{"persona":"critico","persona_name":"El Crítico","ai_generated":true,"verdict":"A ground-up rewrite shipped in July 2026, three releases followed in eight weeks, and the roadmap still lists multi-location file changes as in progress; File Guard is the part done right.","scores":{"reliability":5,"usefulness":6,"cost":7,"longevity":6},"evidence":["https://github.com/agentscope-ai/QwenPaw"]},{"persona":"profesor","persona_name":"El Profesor","ai_generated":true,"verdict":"Scroll Context persists every turn and indexes evicted ones for recall instead of summarising, memory has three layers ending in Markdown, and the Flash models ship with no published evaluation.","scores":{"reliability":7,"usefulness":6,"cost":6,"longevity":7},"evidence":["https://github.com/agentscope-ai/QwenPaw","https://qwenpaw.agentscope.io/"]},{"persona":"inversora","persona_name":"La Inversora","ai_generated":true,"verdict":"Alibaba's AgentScope team, with one-click deployment to Alibaba Cloud ECS, ModelScope Studio and a free always-on AgentScope Platform: the product is a funnel into DashScope and the cloud.","scores":{"reliability":7,"usefulness":7,"cost":6,"longevity":7},"evidence":["https://github.com/agentscope-ai/QwenPaw","https://qwenpaw.agentscope.io/"]},{"persona":"jefa","persona_name":"La Jefa","ai_generated":true,"verdict":"Kernel-level sandboxing on all three desktop OSes is the strongest security page on this board, a multi-user Hub shipped today, SSO is absent, and init --defaults accepts telemetry silently.","scores":{"reliability":5,"usefulness":6,"cost":7,"longevity":6},"evidence":["https://github.com/agentscope-ai/QwenPaw","https://qwenpaw.agentscope.io/"]},{"persona":"hacker","persona_name":"El Hacker","ai_generated":true,"verdict":"Apache-2.0, Ollama at 32k context or LM Studio, a YAML Tool Guard from STRICT to OFF that inspects every call, MCP plus A2A plus ACP drivers, and a persona I edit as SOUL and PROFILE files.","scores":{"reliability":8,"usefulness":8,"cost":9,"longevity":7},"evidence":["https://github.com/agentscope-ai/QwenPaw","https://qwenpaw.agentscope.io/"]},{"persona":"comediante","persona_name":"El Comediante","ai_generated":true,"verdict":"The cloud quick start includes a reminder to set the Studio to non-public, so strangers cannot control your assistant.","scores":null,"evidence":[]}]}