docker-swarm
Audited by Gen Agent Trust Hub on Feb 13, 2026
================================================================================
🟡 VERDICT: MEDIUM
This skill provides comprehensive instructions and a script for managing Docker Swarm. The primary concern is the scripts/swarm-init.sh file, which takes an --advertise-addr argument and passes it directly to the docker swarm init command. While the argument is double-quoted, a sophisticated attacker could craft input that injects additional valid Docker arguments, potentially altering the intended behavior of the docker swarm init command. This represents a limited form of command injection. Additionally, the skill's assets/swarm-stack.yaml references Docker images from external sources, including a custom myapp/api:latest image, which cannot be verified by this analysis. While standard for Docker, it means the skill relies on external, potentially untrusted, code.
Total Findings: 2
🟡 MEDIUM Findings: • Argument Injection in swarm-init.sh
- scripts/swarm-init.sh:26: docker swarm init --advertise-addr "$ADVERTISE_ADDR" The script passes the user-provided ADVERTISE_ADDR directly to 'docker swarm init'. While double-quoted, it's possible to inject additional valid Docker arguments (e.g., '--force-new-cluster') that could alter the command's behavior.
🔵 LOW Findings: • External Docker Image Reference
- assets/swarm-stack.yaml:35: image: myapp/api:latest The skill references a custom Docker image 'myapp/api:latest' which is an unverified external dependency. While common in Docker deployments, the source and contents of this image are unknown. Other images like 'nginx:alpine' and 'postgres:15-alpine' are from trusted sources.
================================================================================