agentleFS
imtiazrayhanimtiazrayhan/agentscamp-library/skills/dockerfile-optimizer/SKILL.md

dockerfile-optimizer

Skill1 starsChanged 5 days agoMITView on GitHub

Shrink and harden an existing Dockerfile — multi-stage builds, cache-friendly layer order, a lean pinned base image, a .dockerignore, and a non-root runtime user — without changing what the image runs. Use when an image is too large, builds are slow because the cache never hits, or a scan flags the container running as root.

  • Deletes or force-pushes

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read the current state: Read the Dockerfile, any .dockerignore, and the build context layout.

    line 22
  2. Very specific:

    Choose a lean, pinned base: Prefer a slim or distroless runtime image over a full OS image where the app allows it.

    line 24
  3. Very specific:

    Run as non-root: Create a dedicated user/group and add USER before the final CMD.

    line 28

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Take a Dockerfile that already works and make it smaller, faster to build, and safer to run — without changing what the container actually does. This skill reads the existing Dockerfile and build context, then applies the standard, high-leverage optimizations: multi-stage builds so build tools never ship in the final image, layer ordering that lets the cache hit on unchanged dependencies, a lean and pinned base image, a .dockerignore, and a non-root runtime user.

When to use this skill

  • The image is large (hundreds of MB to gigabytes) and you want it smaller.
  • Builds are slow because a small source change reinstalls all dependencies (cache never hits).
  • A scanner or reviewer flagged the container running as root, an unpinned :latest base, or secrets in a layer.
  • You're preparing an image for production and want it lean and reproducible.

[!NOTE] This is behavior-preserving. The optimized image must run the same command, expose the same ports, and produce the same result. Don't change the app's runtime behavior, upgrade its language version, or swap frameworks — only how the image is built and packaged.

Instructions

  1. Read the current state. Read the Dockerfile, any .dockerignore, and the build context layout. Identify the language/runtime, the build steps, and the final CMD/ENTRYPOINT. Build once to get a baseline: docker build -t img:before . and record the size with docker images img:before.
  2. Introduce (or tighten) a multi-stage build. Put compilation, dev dependencies, and toolchains in a builder stage; copy only the produced artifacts (binary, dist/, wheels, node_modules for prod) into a minimal final stage. The final image should contain the runtime and the app — not compilers or package caches.
  3. Choose a lean, pinned base. Prefer a slim or distroless runtime image over a full OS image where the app allows it. Pin to a specific tag (and ideally a digest) — never :latest. Match the base to the deployment target's architecture.
  4. Order layers for cache hits. Copy dependency manifests first, install dependencies, then copy application source. This keeps the expensive install layer cached across ordinary code changes. Use BuildKit cache mounts (RUN --mount=type=cache,...) for package-manager caches where supported.
  5. Keep layers clean. Combine related RUN steps, and in the same layer remove what you added — clear apt/apk lists (rm -rf /var/lib/apt/lists/*), don't --no-install-recommends-forget, and avoid leaving package caches. Never COPY secrets or .git into a layer; use build secrets or runtime env instead.
  6. Add a .dockerignore. Exclude .git, node_modules (when installed inside the image), build output, test fixtures, local env files, and docs. A smaller build context means faster builds and no accidental secret leakage.
  7. Run as non-root. Create a dedicated user/group and add USER before the final CMD. Ensure the app's files and any writable dirs are owned appropriately. Add a HEALTHCHECK if the platform uses it.
  8. Verify. Build the optimized image (docker build -t img:after .), confirm it runs and behaves identically, and compare sizes. Report the before/after image size and build-time change with real numbers. If the project uses a scanner (Docker Scout, Trivy, Grype), run it and note the delta.

Examples

A Node service that copied everything before installing, ran as root, and shipped the full build toolchain:

# syntax=docker/dockerfile:1
# ---- builder ----
FROM node:22-bookworm-slim AS builder
WORKDIR /app
COPY package.json package-lock.json ./
RUN --mount=type=cache,target=/root/.npm npm ci
COPY . .
RUN npm run build && npm prune --omit=dev

# ---- runtime ----
FROM node:22-bookworm-slim
WORKDIR /app
ENV NODE_ENV=production
COPY --from=builder /app/node_modules ./node_modules
COPY --from=builder /app/dist ./dist
USER node
EXPOSE 3000
CMD ["node", "dist/server.js"]

Report the outcome: image size before → after, whether the build cache now survives a source-only change, the non-root user added, and any scanner findings resolved. Note anything you intentionally left alone (e.g. a base image the app pins for a native dependency).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.