Intel RealSense ROS2 MCP Server — 封装 realsense2_camera 的 ROS2 接口,提供节点 launch 管理、topic 订阅捕帧、参数控制、TF 查询和多相机协同。
Server has 11 tools with complete input schemas and descriptions in Chinese. Naming follows verb_noun pattern (launch_camera, stop_camera, list_topics, capture_*). However, descriptions lack English translations and are moderately detailed (avg ~80 chars). Parameter descriptions are present but terse. Output schemas are not documented, tools return JSON strings without specifying field structure. Error handling returns wrapped JSON but lacks recovery guidance or error classification. No tool annotations (readOnlyHint/destructiveHint). Safety validation exists (serial, camera_name, resolution, fps, timeout, file_path) but is not exposed in schema constraints.
从 ROS2 color topic 订阅一帧彩色图像并保存为 PNG。
从 ROS2 depth topic 订阅一帧深度图像并保存。
从 ROS2 infrared topic 捕获一帧红外图像。
捕获对齐的 RGBD 图像对(颜色+深度)并分别保存。
获取 IMU 数据(方向、角速度、线加速度)。
测量 ROS2 topic 的发布频率。
获取指定 ROS2 topic 的类型和发布者信息。
启动一个 realsense2_camera ROS2 节点。
Output schemas not documented. Tools return JSON strings but field structure is not specified. LLMs cannot plan downstream calls or extract required data without seeing response schema.
Descriptions are in Chinese only and lack English translations. LLMs trained primarily on English may misinterpret intent. Descriptions also lack WHEN to use guidance and prerequisites.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 57 | <=2025-11-25 | v2 |
列出所有由本 bridge 管理的活动 ROS2 相机节点进程。
列出 ROS2 topics,可按相机名过滤。
停止指定的 realsense2_camera ROS2 节点。
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). LLMs cannot distinguish safe read-only tools from destructive ones (launch_camera, stop_camera, capture_* are WRITE; list_*, get_* are READ_ONLY).
Error responses lack recovery guidance. _err() returns bare error message without suggesting next steps or error classification (retryable vs user-fixable vs fatal).
Parameter constraints (resolution 320-1920, fps 1-90, timeout 0.1-30) are validated in code but not exposed in JSON Schema. LLMs cannot see constraints and may pass invalid values.