create-multistage-dockerfile
About
This Claude skill generates optimized multi-stage Dockerfiles that separate build and runtime environments to create minimal production images. It's ideal when your Docker images are too large, contain unnecessary build tools, or need to run in constrained environments like serverless platforms. The skill covers techniques including artifact copying between stages and using minimal base images like Alpine, Distroless, or scratch.
Quick Install
Claude Code
Recommendednpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/create-multistage-dockerfileCopy and paste this command in Claude Code to install this skill
Documentation
Create Multi-Stage Dockerfile
Build multi-stage Dockerfiles producing minimal production images. Separate build tooling from runtime.
When Use
- Production images too large (>500MB for compiled languages)
- Build tools (compilers, dev headers) in final image
- Need separate images for dev and prod from one Dockerfile
- Deploying to constrained environments (edge, serverless)
Inputs
- Required: Existing Dockerfile or project to containerize
- Required: Language and build system (npm, pip, go build, cargo, maven)
- Optional: Target runtime base (slim, alpine, distroless, scratch)
- Optional: Size budget for final image
Steps
Step 1: Identify Build vs Runtime Dependencies
| Category | Build Stage | Runtime Stage |
|---|---|---|
| Compilers | gcc, g++, rustc | Not needed |
| Package managers | npm, pip, cargo | Sometimes (interpreted langs) |
| Dev headers | -dev packages | Not needed |
| Source code | Full source tree | Only compiled output |
| Test frameworks | jest, pytest | Not needed |
Step 2: Structure the Multi-Stage Build
Core pattern: build in fat image, copy artifacts to slim image.
# ---- Build Stage ----
FROM <build-image> AS builder
WORKDIR /src
COPY <dependency-manifest> .
RUN <install-dependencies>
COPY . .
RUN <build-command>
# ---- Runtime Stage ----
FROM <runtime-image>
COPY --from=builder /src/<artifact> /<dest>
EXPOSE <port>
CMD [<entrypoint>]
Step 3: Apply Language-Specific Patterns
Node.js (pruned node_modules)
FROM node:22-bookworm AS builder
WORKDIR /src
COPY package.json package-lock.json ./
RUN npm ci
COPY . .
RUN npm run build && npm prune --omit=dev
FROM node:22-bookworm-slim
RUN groupadd -r app && useradd -r -g app app
WORKDIR /app
COPY --from=builder /src/dist ./dist
COPY --from=builder /src/node_modules ./node_modules
COPY --from=builder /src/package.json .
USER app
EXPOSE 3000
CMD ["node", "dist/index.js"]
Python (virtualenv copy)
FROM python:3.12-bookworm AS builder
WORKDIR /src
RUN python -m venv /opt/venv
ENV PATH="/opt/venv/bin:$PATH"
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
COPY . .
FROM python:3.12-slim-bookworm
COPY --from=builder /opt/venv /opt/venv
ENV PATH="/opt/venv/bin:$PATH"
WORKDIR /app
COPY --from=builder /src .
RUN groupadd -r app && useradd -r -g app app
USER app
EXPOSE 8000
CMD ["python", "app.py"]
Go (static binary to scratch)
FROM golang:1.23-bookworm AS builder
WORKDIR /src
COPY go.mod go.sum ./
RUN go mod download
COPY . .
RUN CGO_ENABLED=0 GOOS=linux go build -ldflags="-s -w" -o /server ./cmd/server
FROM scratch
COPY --from=builder /etc/ssl/certs/ca-certificates.crt /etc/ssl/certs/
COPY --from=builder /server /server
EXPOSE 8080
ENTRYPOINT ["/server"]
Rust (static musl binary)
FROM rust:1.82-bookworm AS builder
RUN apt-get update && apt-get install -y musl-tools && rm -rf /var/lib/apt/lists/*
RUN rustup target add x86_64-unknown-linux-musl
WORKDIR /src
COPY Cargo.toml Cargo.lock ./
RUN mkdir src && echo "fn main() {}" > src/main.rs \
&& cargo build --release --target x86_64-unknown-linux-musl \
&& rm -rf src
COPY . .
RUN touch src/main.rs && cargo build --release --target x86_64-unknown-linux-musl
FROM scratch
COPY --from=builder /src/target/x86_64-unknown-linux-musl/release/myapp /myapp
EXPOSE 8080
ENTRYPOINT ["/myapp"]
Got: Final image has only runtime and compiled artifacts.
If fail: Check COPY --from=builder paths. Use docker build --target builder to debug build stage.
Step 4: Choose Runtime Base
| Base | Size | Shell | Use Case |
|---|---|---|---|
scratch | 0 MB | No | Static Go/Rust binaries |
gcr.io/distroless/static | ~2 MB | No | Static binaries + CA certs |
gcr.io/distroless/base | ~20 MB | No | Dynamic binaries (libc) |
*-slim | 50-150 MB | Yes | Interpreted languages |
alpine | ~7 MB | Yes | When shell access needed |
Note: Alpine uses musl libc. Some Python wheels and Node native modules may not work. Prefer -slim (glibc) for interpreted languages.
Step 5: Build Args Across Stages
ARG APP_VERSION=0.0.0
FROM golang:1.23 AS builder
ARG APP_VERSION
RUN go build -ldflags="-X main.version=${APP_VERSION}" -o /server .
FROM gcr.io/distroless/static
COPY --from=builder /server /server
ENTRYPOINT ["/server"]
Build with: docker build --build-arg APP_VERSION=1.2.3 .
Note: ARG before FROM is global. Each stage must re-declare ARG to use it.
Step 6: Compare Image Sizes
# Build both variants
docker build -t myapp:fat --target builder .
docker build -t myapp:slim .
# Compare sizes
docker images --format "table {{.Repository}}\t{{.Tag}}\t{{.Size}}" | grep myapp
Got: Production image 50-90% smaller than build stage.
Checks
-
docker buildfinishes for all stages - Final image has no build tools (compilers, dev headers)
-
docker runworks from slim image - Image size significantly smaller vs single-stage
-
COPY --from=builderpaths right - No source code leaks into production image
Pitfalls
- Missing runtime libraries: Compiled code may need shared libraries (
libc,libssl). Test slim image thoroughly. - Broken
COPY --frompaths: Artifact path must match exactly. Usedocker build --target builderthendocker run --rm builder ls /pathto debug. - Alpine musl issues: Native Node.js addons and some Python packages fail on Alpine. Use
-sliminstead. - Global ARG scope:
ARGdeclared beforeFROMis available toFROMlines only. Re-declare inside each stage that needs it. - Forgetting CA certificates:
scratchhas no certificates. Copy/etc/ssl/certs/ca-certificates.crtfrom builder or use distroless.
See Also
create-dockerfile- single-stage general Dockerfilescreate-r-dockerfile- R-specific Dockerfiles with rocker imagesoptimize-docker-build-cache- layer caching and BuildKit featuressetup-compose-stack- compose configurations using multi-stage images
GitHub Repository
Related Skills
content-collections
MetaThis skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.
polymarket
MetaThis skill enables developers to build applications with the Polymarket prediction markets platform, including API integration for trading and market data. It also provides real-time data streaming via WebSocket to monitor live trades and market activity. Use it for implementing trading strategies or creating tools that process live market updates.
creating-opencode-plugins
MetaThis skill helps developers create OpenCode plugins that hook into 25+ event types like commands, files, and LSP operations. It provides the plugin structure, event API specifications, and implementation patterns for JavaScript/TypeScript modules. Use it when you need to intercept, monitor, or extend the OpenCode AI assistant's lifecycle with custom event-driven logic.
sglang
MetaSGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.
