{"data":{"kind":"file","path":"README.md","version_id":"frhlstizt79o1dynloi7uw16","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":2034,"modified_at":"2026-08-05T12:56:44.134000","content_hash":"8d22ed012bd47df09730094b33a99690d775dd583a0fdd81912b2485bc910c9e"},"entries":[],"content":"# Docker / Container Manager\n\nAn evaluation environment that tests an LLM's ability to work with Docker and container technologies across four difficulty levels.\n\n## What It Tests\n\n**Level 0 — Basics:**\n- Writing simple Dockerfiles (Python/Flask)\n- Basic docker-compose.yml (single service + volume)\n- Docker CLI commands\n- .dockerignore files\n\n**Level 1 — Intermediate:**\n- Multi-stage Dockerfiles (Go, Java)\n- Multi-service compose (React + API + DB with health checks)\n- Image optimization techniques\n- Non-root user configurations\n\n**Level 2 — Advanced:**\n- Production-ready compose with nginx, replicas, resource limits\n- Secure Rust builds with distroless\n- Network isolation (frontend/backend/database)\n- BuildKit cache mounts\n\n**Level 3 — Expert:**\n- Kubernetes deployment YAML (Deployment + Service + HPA)\n- Security hardening (tini, dropped capabilities, read-only fs)\n- CI/CD pipelines with Docker (GitHub Actions + Trivy)\n- ELK stack logging with Filebeat sidecars\n- Development environments with hot-reload\n- Network troubleshooting guides\n\n## Scoring Rubric\n\nEach response is scored on four dimensions:\n\n| Metric | Weight | Description |\n|--------|--------|-------------|\n| Keyword Coverage | 35% | Presence of expected Docker keywords, syntax, and constructs |\n| Structure Score | 15% | Code blocks, organization, explanation quality |\n| Best Practices | 20% | Alpine/slim images, cache cleanup, health checks, non-root, multi-stage |\n| Accuracy | 30% | Overlap with reference solution tokens |\n\n## Usage\n\n```python\nimport verifiers as vf\n\nenv = vf.load_environment(\"docker-container-manager\")\n\n# Evaluate at a specific difficulty level\nenv = vf.load_environment(\"docker-container-manager\", level=2)\n```\n\n## Dataset\n\n20 tasks across 4 difficulty levels covering the full Docker/container workflow:\n- Dockerfile authoring and optimization\n- docker-compose multi-service orchestration\n- Networking, volumes, and security\n- Kubernetes manifests\n- CI/CD integration\n- Logging and monitoring stacks\n\n## License\n\nMIT\n","encoding":"utf-8","truncated":false,"total_bytes":2034},"status":null}