aws-transform

SUSPICIOUS: the skill's overall purpose is coherent for AWS-based code transformation, and core network endpoints align with AWS ownership, so this is not confirmed malware. However, its footprint is broad: raw installer execution, mutable external code, trust-all-tools execution, IAM policy attachment, remote infrastructure deployment, secrets handling, and forwarding local config into remote jobs make it disproportionately risky for an AI skill.

This fragment appears to implement legitimate AWS runtime IAM provisioning, but it performs high-impact persistent privilege changes. The main risks are broad permissions, automatic attachment to the caller, and acceptance of any pre-existing policy with the same name without document validation. No clear malware or data-theft behavior is present in the shown code; review the policy JSON, execution context, trust boundaries, and whether caller-controlled inputs can influence the policy path or ARN.

The fragment describes a legitimate remote AWS transformation workflow, and it contains no direct evidence of malware or intentional data exfiltration. However, it presents substantial supply-chain and operational security risk because it clones an unpinned public branch, executes setup and Dockerfile commands without verification, uses a mutable public ECR latest image, forwards arbitrary commands to remote containers, and handles high-value credentials. The deployment and IAM changes should be reviewed, dependencies should be pinned and verified, and job isolation and least-privilege permissions should be confirmed before use.