Dockerfile And Image Build
Docker (Engine, Hub, Build Cloud, Scout, Desktop, Model Runner) · Docker
Container Platform — Docker
Evaluates Docker's Dockerfile & Image Build across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Container Platform eval coverage.
About Docker
Docker is the container platform — Docker Engine, Docker Desktop, Docker Hub registry, Docker Build Cloud for managed cloud builders, Docker Scout for image vulnerability scanning and supply-chain policy, Docker Compose for multi-container dev, and Docker Model Runner for local LLM inference. Millions of developers and tens of thousands of enterprises ship containerized software with Docker.
Sample tests· showing 3 of 9
| # | Input | Expected behavior | Check |
|---|---|---|---|
| 01 | Agent writes a Dockerfile with a single FROM golang:1.22 stage that runs 'go build' then COPYs the binary into /app and sets CMD. Final image is 1.2 GB containing the entire Go toolchain in production. | Use a multi-stage build: FROM golang:1.22 AS build (compile), then FROM gcr.io/distroless/static-debian12 or scratch as the final stage, COPY --from=build /out/app /app. Final image contains only the binary and minimal runtime. Verify via 'docker image ls' size delta. | Pass / FailAi Platformhigh |
| 02 | Agent needs an NPM_TOKEN during 'npm install' but not in the final image. Current Dockerfile uses ARG NPM_TOKEN and ENV NPM_TOKEN=$NPM_TOKEN. | Use BuildKit secret mount: RUN --mount=type=secret,id=npm,target=/root/.npmrc npm ci. Pass --secret id=npm,src=$HOME/.npmrc at build time. Token never lands in a layer and 'docker history' shows no secret. ARG values are visible in 'docker history' and ENV values persist in the final image. | Pass / FailAi Platformcritical |
| 03 | Python project rebuilds reinstall the same 200 wheels on every layer change because pip's cache lives in a layer that's invalidated by source edits. | Add RUN --mount=type=cache,target=/root/.cache/pip pip install -r requirements.txt. The cache mount persists across builds outside any layer, so reinstalls hit the wheel cache even when surrounding layers change. Requires BuildKit (# syntax=docker/dockerfile:1). | Pass / FailAi Platformmedium |
How this eval is graded
Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
Rubric criteria
- Docker
- Ai Platform
- Dockerfile And Image Build
Recommended for
Works with
Related evals
Claude API
Evaluates Anthropic's Batch API across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
View AI PlatformClaude API
Evaluates Anthropic's Extended Thinking across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
View AI PlatformClaude API
Evaluates Anthropic's Files API & Citations across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
ViewFrequently asked questions
What does the Dockerfile And Image Build eval for Docker Docker (Engine, Hub, Build Cloud, Scout, Desktop, Model Runner) test?+
Evaluates Docker's Dockerfile & Image Build across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Container Platform eval coverage.
How is the Dockerfile And Image Build eval scored?+
The judge rubric: Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
How many test cases does this eval pack include?+
The Dockerfile And Image Build pack for Docker Docker (Engine, Hub, Build Cloud, Scout, Desktop, Model Runner) contains 9 test cases. 3 sample cases are shown free on this page; the full set runs in a Corsac workspace.
How do I run this eval?+
Sign up for Corsac, connect your model or agent endpoint, and run the Dockerfile And Image Build pack as-is or after customizing thresholds. Results land in your workspace with per-case scores, and you can gate releases on the pack in CI via the REST API.
Run this eval in your workspace
Connect your data, configure thresholds, and review results with your team.