io.github.Agentled/mcp-server

The materials show a required AGENTLED_API_KEY, which is a sensitive credential; if the server or logs are handled improperly, it could be abused to access the related service. No README details are provided on key storage, least privilege, or rotation practices.

No remote host is listed in the objective checks, but the tool requires a cloud API key, so it likely communicates with Agentled-related services. Its stated workflow orchestration, long-term memory, and 100+ integrations suggest user inputs or workflow data may be sent to external services; endpoint, data-type, and retention details are not documented.

The system explicitly flags it as executes-code, indicating local code or process execution capability; this is a normal but high-privilege trait for MCP tools and should be used carefully. The materials do not describe execution boundaries, sandboxing, command allowlists, or privilege isolation.

The mention of long-term memory and many integrations implies potential access to and persistence of workflow context or external resource data, but with no README it is unclear what local files, caches, or third-party resources are actually read or written. There is no evidence of clearly excessive permissions beyond its stated function, so this is caution rather than high risk.

It comes from an official registry, has a public source repository, and was updated within the last year, providing auditability and some maintenance signals. While the low star count and undeclared license add adoption and compliance uncertainty, they are not by themselves enough to raise this to high risk.