OpenClaw MCP Server

The materials explicitly state that no keys or environment variables are required, and there is no indication that users must provide API tokens, account credentials, or other sensitive authentication data, so credential exposure appears low.

The description says it exposes the OpenClaw assistant to Claude Desktop via Streamable HTTP transport, which implies HTTP-based communication; however, the materials also declare no remote endpoints, and there is no evidence of data being sent to undeclared third-party services. Because network interaction is part of its functionality, verify that actual traffic is limited to local or expected client endpoints.

The system checks explicitly mark it as having executes-code capability, meaning the MCP tool can run code or processes on the local machine. This is a common powerful capability for this class of tools and not by itself a high-risk red flag, but it should be run with least privilege and isolation.

The materials do not specify the exact scope of readable or writable data; combined with its code-execution capability, it could in principle access local data available to the host process. There is no evidence that it requests data permissions clearly beyond its stated function, but the lack of documentation limits visibility.

Positive factors include that the project is open source, auditable, and MIT licensed; however, it comes from a third-party registry, has only 0 GitHub stars, unknown maintenance status, and no README, so while basic transparency exists, trust signals are weak. Review the source code and dependency list before use.