- Reliability 5/20
- The manifest is well-formed and superficially consistent (name, npm package, version, stdio transport), and the README claims structured JSON output for Docker tools. However, no source code, tests, or CI workflow for this specific server was provided, so startup, MCP handshake, and actual tool behavior cannot be verified. Under the static review rules, missing evidence is a deduction; the score stays well below 12.
- Security and permissions 5/20
- No red-line issues (malware, credential theft, exfiltration, unconfirmed destructive defaults) are visible in the supplied material. Yet there is no source or config showing credential handling, least privilege, confirmation for side-effectful Docker operations (build, compose, images), or data-flow disclosure. The documented tool-filtering env vars reduce exposure but are not a substitute for built-in confirmation, so the score is low.
- Maintenance 6/20
- The repository declares an MIT license, version 0.22.0, and CI/OpenSSF badges, and is not archived. These are repository-level signals from untrusted README/manifest evidence; there is no commit history, release notes, dependency-update evidence, or security-response channel tied to the Docker server. Sustained maintenance cannot be confirmed.
- Documentation 6/20
- The README provides general installation, tool-selection env vars, troubleshooting, and a git MCP config example, which are useful. But Docker-specific tool parameters, response schemas, auth/limits are only linked and not supplied; the examples are for git/npm rather than Docker, so the documentation for this server is incomplete.
- Setup experience 7/20
- Generic setup via npx @paretools/init and client config examples exist, and the Docker package identifier plus stdio transport make a straightforward npx-based configuration plausible. However, no Docker-specific config example or verifiable execution evidence was provided; per the static cap of 15 and the uncertainty involved, the score is modest.