RoboParts 机器人零部件兼容性

What RoboParts 机器人零部件兼容性 MCP server does

RoboParts 机器人零部件兼容性 MCP server provides an agent-facing way to work with the RoboParts open data layer for humanoid robotics. The underlying dataset covers actuators, chips, sensors, protocols, data-acquisition systems, robot AI models, interfaces, platforms, grippers, flexible actuators, reducers, controllers, and other component categories.

The project is vendor-neutral and is intended for cross-manufacturer interface and compatibility research. Records can include source, confidence, and last-verified metadata. The dataset also records conformance or compatibility status for standards and ecosystems such as GB modular robotics requirements, ISO 8373, IEC 61508, and ROS2 when those fields are available.

How it works

The service is backed by RoboParts data and API endpoints hosted at https://roboparts.cc. Agents can work from the complete entity collection, query the compatibility matrix, retrieve a specific entity, or search by keyword. Search accepts optional category and result-limit parameters, with an option to include quarantined records.

Compatibility is represented across four dimensions: electrical, mechanical, protocol, and software. The repository also describes a five-factor selection score using torque, speed, precision, weight, and cost. These functions are useful for narrowing candidates, but they do not replace engineering validation or vendor confirmation.

Tools and capabilities

The documented API surface includes:

• GET /api/entities.json for the full entity list.
• GET /api/compatibility_matrix.json for compatibility relationships.
• GET /api/entity/{id} for one entity’s details, including fields that may be marked as paid or restricted.
• GET /api/search?q=keyword for keyword search, with optional category, limit, and include_quarantine parameters.

RoboParts 机器人零部件兼容性 MCP server can therefore support tasks such as finding candidate actuators, comparing sensor or controller records, locating protocol entries, checking whether an interface claim exists, and identifying documented domestic-substitution relationships. The repository’s categories also cover robotic AI models, large language models, data-collection equipment, and complete robot platforms, so searches are not limited to physical parts.

Limitations and notes

The repository reports 802 entities in the current README statistics, including physical components, interface specifications, AI model software, companies, and market intelligence. Counts and category totals are generated from api/entities.json and may change as the data is updated.

Mechanical interface coverage is explicitly incomplete: the README reports declarations for only 25 of 439 applicable records, or 5.69 percent. A missing compatibility result should therefore not be treated as proof that two parts cannot connect. The project requires sources for accepted data and prefers not_declared over an unsupported guess.

The records contain vendor-published values and have not been independently reproduced by the project. The README states that there are currently no cross-vendor entries with an A-level rating for direct comparison. Use source tier and confidence fields when deciding whether a result is suitable for design or procurement.

Licensing is split. Code is MIT-licensed, while the data in the API and related dataset locations is released under CC BY 4.0. The README states that API tokens should be managed through environment variables, but it does not specify an environment-variable name or provide MCP installation commands in the supplied material.