MCP Codex Worker

The materials explicitly state that no keys or environment variables are required. No API tokens, account credentials, or other sensitive secrets are requested, so credential exposure risk appears low.

No remote endpoints or egress hosts are declared in the materials. Although the description mentions using cheaper AI models, the provided materials do not specify any network destination or data exfiltration path, so no clear egress red flag is shown.

The system checks explicitly indicate executes-code, and the description includes an async worker, patching, and testing, indicating normal local code execution and test-running capability. This is consistent with the tool's purpose, but it should be run in an isolated environment with a limited working scope.

The description says it can perform bulk code reading, patching, and testing, which implies access to project source code and possible file modifications. This is a normal level of data access for its stated function, but the materials do not define finer-grained permission boundaries.

Positive factors are that it is open source under the MIT license and can be audited. However, it comes from a third-party registry, has only 0 GitHub stars, no README, and unknown maintenance status, so trust is limited; review the source and dependencies before using it in sensitive environments.