contextleak

Your .gitignore won't save your Docker image. contextleak checks whether the sensitive files sitting in your build context — .env, private keys, credentials, cloud configs — would actually be COPY'd/ADD'd into your image, by simulating .dockerignore matching against every COPY/ADD instruction in your Dockerfile.

pip install contextleak
cd your-project
contextleak

Scanned 214 files in build context (12 excluded by .dockerignore), 4 COPY/ADD instruction(s).

LEAK: 2 instruction(s) would copy sensitive files:

  [COPY line 8] COPY . /app  (stage: stage-1)
    - .env  (environment file, usually holds secrets)
    - certs/server.key  (private key material)

  [ADD line 11] ADD config/ /etc/myapp/
    - config/credentials.json  (file name suggests stored credentials)

Exit code 1 when a leak is found, 0 when clean — drop it into CI.

Install

pip install contextleak

Requires Python 3.9+. Zero dependencies (stdlib only).

Usage

contextleak [context] [-f DOCKERFILE] [--ignore-file IGNORE] [--json]

• context — build context directory (default: current directory).
• -f, --dockerfile — Dockerfile path (default: <context>/Dockerfile).
• --ignore-file — ignore file path (default: <context>/.dockerignore).
• --json — machine-readable output for CI pipelines.

What it does:

1. Parses the Dockerfile's COPY/ADD instructions (multi-stage aware; COPY --from=<stage> copies come from another stage, not the context, so they are skipped).
2. Simulates .dockerignore matching with Docker's own semantics — patterns anchored to the context root, ** crossing directories, ! negation with last-match-wins, directory patterns excluding their contents.
3. Expands each COPY/ADD source glob against the files that would actually be sent to the daemon.
4. Flags sensitive names: .env/.env.*, *.pem, *.key, SSH private keys, *credential*, *secret*, *token*, .npmrc, .pypirc, Terraform state and *.tfvars, kubeconfigs, service-account files, and the .git directory.

It also warns when an instruction references files that .dockerignore (or absence) removed — that's a build that fails or silently copies nothing.

Differentiation

Tools like Trivy scan the finished image for known vulnerabilities (CVEs in packages). contextleak checks a different, earlier layer: the build context. A secret doesn't need a CVE to hurt you — if your .env is COPY'd into the image at build time, it ships with every push of that image, and no vulnerability scanner will flag it because it's your own file, not a known-vulnerable package. One cut, one job: does your build context leak secrets into the image? Nothing else.

License

MIT — see LICENSE.