VectorClaw MCP

VectorClaw MCP

VectorClaw connects OpenClaw to an Anki / Digital Dream Labs Vector ro机器人 through MCP. It provides practical ro机器人 control primitives for speech, movement, camera capture, and 状态/sensor reads.

What you can do Speak text with vector_say Move and position with vector_drive, vector_head, vector_lift Capture camera images with vector_look and vector_capture_image Read ro机器人 状态 with vector_状态, vector_pose, vector_proximity_状态, vector_touch_状态 Build look → reason → act 工作流s Vision requirement for look → reason → act

For see → reason → act 工作流s, the 代理 must either be vision-capable itself (e.g., a VLM) or have 访问 to a separate vision 模型/image-interpretation 工具 to analyze camera images before choosing actions.

Requirements Vector ro机器人 配置d and reachable Wire-Pod 运行ning SDK 配置d at ~/.anki_vector/sdk_config.ini VECTOR_SERIAL 环境 variable 设置 Quick 设置up 安装 package: pip 安装 vectorclaw-mcp 配置 SDK: python3 -m anki_vector.配置 导出 ro机器人 serial: 导出 VECTOR_SERIAL=your-serial 添加 MCP server: { "mcpServers": { "vectorclaw": { "command": "python3", "args": ["-m", "vectorclaw_mcp.server"], "env": { "VECTOR_SERIAL": "${VECTOR_SERIAL}" } } } }

工具 coverage

Hardware-verified core 工具s vector_say, vector_drive_off_charger, vector_drive, vector_emergency_停止, vector_head, vector_lift, vector_look, vector_capture_image, vector_face, vector_扫描, vector_vision_re设置, vector_pose, vector_状态, vector_charger_状态, vector_touch_状态, vector_proximity_状态

Experimental 工具s vector_animate, vector_drive_on_charger, vector_find_faces, vector_列出_visible_faces, vector_face_检测ion, vector_列出_visible_objects, vector_cube

Current limitations Charger return (vector_drive_on_charger) is currently unreliable Face/object 检测ion is currently inconsistent Visual interpretation requires the vision capability described above Documentation MCP API: docs/MCP_API_REFERENCE.md SDK Reference: docs/VECTOR_SDK_REFERENCE.md Hardware 记录: docs/HARDWARE_SMOKE_记录.md Repo: https://github.com/danmartinez78/VectorClaw