io.github.CSOAI-ORG/agent-commerce-payments-mcp

The materials indicate no required keys or environment variables, and there is no evidence that API tokens, account passwords, or other sensitive credentials are requested; this keeps credential exposure relatively low. Still, risk would change if later versions introduce authentication.

It is known to contact the remote endpoint api.meok.ai, which is a standard network-egress capability. Because the README is absent, it is unclear what user input, payment data, or metadata may be transmitted, but there is no concrete red flag showing exfiltration to unrelated or unknown endpoints.

System checks show this tool has executes-code capability, meaning it may start local processes or execute code; this is a common but high-impact MCP capability and should be granted carefully. The available materials do not show it requesting system privileges clearly beyond its stated purpose.

The provided materials do not specify what local files, payment records, or other resources it can read or write, so its data-access scope is unclear. As an MCP tool with execution capability, it should be run with least privilege by default, but there is currently no evidence of explicit overbroad access.

The source is listed in an official registry, is open source, and has been updated within the last year, all of which are positive signals that lower risk. On the other hand, the missing README, undeclared license, and low community adoption (0 stars) limit assessability and maturity, making caution more appropriate than high risk.