twinmind-deploy-integration

SUSPICIOUS. The skill’s network targets are vendor-aligned and there is no obvious exfiltration or malicious installer, but the documented API-key flow is not supported by the official TwinMind MCP documentation I found, and the skill is third-party-authored with broader shell permissions than necessary. This looks more like an unreliable or poorly scoped integration guide than confirmed malware.

The supplied infrastructure configuration does not contain evident malware, credential theft, data exfiltration, or sabotage. It has moderate supply-chain and deployment security weaknesses: mutable `latest` tags, a mismatch between pushed and deployed image tags, publicly mapped unauthenticated Redis, and incomplete evidence of least-privilege IAM, artifact signing, and scanning. Pin images by immutable digest or commit tag, restrict Redis networking and enable authentication where appropriate, define workflow permissions explicitly, and verify that the deployment references the built artifact.