Build Yaml Misconfiguration logo

Build Yaml Misconfiguration

Organization
RedHatProductSecurity
build-yaml-misconfiguration

Apply when reviewing or writing GitLab CI (.gitlab-ci.yml), Tekton pipeline/task YAML, or Containerfiles/Dockerfiles used in CI builds. Detects common misconfigurations that expose secrets, weaken isolation, skip security gates, grant excessive privileges, or produce insecure container images.

Overview

PublisherRedHatProductSecurity
Repositoryprodsec-skills
Skill namebuild-yaml-misconfiguration
Stars
56
Forks
9
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by RedHatProductSecurity on GitHub. Read the source before you install it.

Installation

Install the Build Yaml Misconfiguration AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/RedHatProductSecurity/prodsec-skills/tree/main/module/skills/build-yaml-misconfiguration
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/RedHatProductSecurity/prodsec-skills.git /tmp/prodsec-skills
mkdir -p .claude/skills
cp -r /tmp/prodsec-skills/module/skills/build-yaml-misconfiguration .claude/skills/build-yaml-misconfiguration
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Build Yaml Misconfiguration in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Build Yaml Misconfiguration on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Build Yaml Misconfiguration is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Build YAML Misconfiguration

Detect and prevent security misconfigurations in GitLab CI and Tekton pipeline definitions. Misconfigured build YAML is a common source of secret leakage, privilege escalation, and supply chain compromise.

GitLab CI (.gitlab-ci.yml)

Secret Exposure

  • Never hardcode secrets in .gitlab-ci.yml, variables:, or script: blocks. Use CI/CD protected variables or an external vault.
  • Mask and protect variables: set masked: true and protected: true on sensitive CI/CD variables. Masked variables are redacted from job logs.
  • Never echo secrets: avoid echo $SECRET, printenv, env, or set -x in scripts that handle credentials. Debug output is stored in job logs.
  • Restrict variable scope: use protected: true to limit variables to protected branches/tags only. Use environment scoping to restrict which jobs see which secrets.
  • Artifacts and caches: never include files containing secrets (.env, credentials, tokens) in artifacts: or cache: paths. These are stored and downloadable.
yaml
# BAD -- secret visible in logs and available on all branches
variables:
  DB_PASSWORD: "hunter2"

# GOOD -- use CI/CD protected variable (set in GitLab UI)
# Reference as $DB_PASSWORD in scripts; set masked + protected in settings

Runner and Execution Isolation

  • Never use shared runners for secret-heavy jobs without understanding the trust boundary. Prefer project-specific or group runners for sensitive pipelines.
  • Never use privileged: true in runner config unless absolutely required (e.g., container builds). Privileged runners can escape to the host.
  • Use docker or kubernetes executor over shell executor. Shell executors share the host filesystem and can leak state between jobs.
  • Tag sensitive jobs to run only on hardened, dedicated runners.
yaml
# BAD -- no runner restriction, runs on any available shared runner
deploy-production:
  script:
    - ./deploy.sh

# GOOD -- pinned to a dedicated runner
deploy-production:
  tags:
    - production-deploy
    - privileged-denied
  script:
    - ./deploy.sh

Pipeline Permissions and Trust

  • Protected branches: ensure deploy and release jobs use only: refs: [main] or rules: with branch restrictions. Never allow feature branches to trigger production deploys.
  • Merge request pipelines: use rules: - if: '$CI_PIPELINE_SOURCE == "merge_request_event"' to prevent MR pipelines from accessing protected variables.
  • allow_failure: true: never set this on security scanning jobs (SAST, SCA, secret detection). Failures in security gates must block the pipeline.
  • when: manual: use for production deploys to enforce human approval. Never auto-deploy to production from CI.
yaml
# BAD -- security scan can fail silently
sast:
  allow_failure: true
  script:
    - run-sast-scanner

# GOOD -- security scan blocks pipeline on failure
sast:
  allow_failure: false
  script:
    - run-sast-scanner
  rules:
    - if: '$CI_PIPELINE_SOURCE == "merge_request_event"'

Container Image Builds in CI

  • Never use docker:dind with --privileged without understanding the risk. Prefer Kaniko, Buildah, or Podman for unprivileged image builds.
  • Pin base images by digest, not just tag. Tags are mutable.
  • Do not mount the Docker socket (/var/run/docker.sock) in CI jobs.
  • Scan built images before pushing to registry.
yaml
# BAD -- privileged DinD with mutable tag
build-image:
  image: docker:latest
  services:
    - docker:dind
  variables:
    DOCKER_HOST: tcp://docker:2375
  script:
    - docker build -t myapp .

# GOOD -- unprivileged build with pinned image
build-image:
  image: quay.io/buildah/stable@sha256:<digest>
  script:
    - buildah bud --layers -t myapp .
    - buildah push myapp $CI_REGISTRY_IMAGE:$CI_COMMIT_SHA

Include and Extends Risks

  • Audit remote includes: include: remote: pulls YAML from external URLs. If the remote changes, your pipeline changes. Pin to a specific ref or commit.
  • Audit cross-project includes: include: project: pulls from other GitLab projects. Ensure the source project has proper access controls.
  • Review inherited configuration: extends: and include: can override variables:, script:, and rules: in unexpected ways. Verify the final merged configuration.
yaml
# RISKY -- remote include with no version pinning
include:
  - remote: 'https://example.com/ci-templates/build.yml'

# BETTER -- pinned project include
include:
  - project: 'my-group/ci-templates'
    ref: 'v2.1.0'
    file: '/templates/build.yml'

Tekton Pipelines

Secret Exposure

  • Never inline secrets in Task or Pipeline YAML. Use Kubernetes Secrets referenced via secretKeyRef in env: or mounted as volumes.
  • Never log secrets in steps: avoid echo, printenv, cat on mounted secret files in script: blocks. Tekton step logs are stored and accessible.
  • Workspace binding: when binding secrets as workspaces, ensure the workspace is not shared with untrusted tasks in the same pipeline.
  • Results: never write secrets to Tekton results. Results are stored in the TaskRun/PipelineRun status and visible to anyone with read access.
yaml
# BAD -- secret hardcoded in task
steps:
  - name: deploy
    image: registry.redhat.io/ubi9/ubi-minimal
    script: |
      curl -H "Authorization: Bearer my-secret-token" https://api.example.com

# GOOD -- secret from Kubernetes Secret
steps:
  - name: deploy
    image: registry.redhat.io/ubi9/ubi-minimal
    env:
      - name: API_TOKEN
        valueFrom:
          secretKeyRef:
            name: deploy-creds
            key: token
    script: |
      curl -H "Authorization: Bearer $API_TOKEN" https://api.example.com

Step Image Security

  • Pin images by digest: never use :latest or mutable tags for step images. A compromised tag can inject malicious code into your pipeline.
  • Use trusted registries: prefer registry.redhat.io or your internal registry. Avoid pulling from untrusted public registries in production pipelines.
  • Minimal images: use UBI minimal or distroless images. Avoid full OS images with unnecessary tools that expand the attack surface.
yaml
# BAD -- mutable tag from public registry
steps:
  - name: build
    image: ubuntu:latest

# GOOD -- pinned digest from trusted registry
steps:
  - name: build
    image: registry.redhat.io/ubi9/ubi-minimal@sha256:<digest>

Task and Pipeline Permissions

  • Run steps as non-root: set securityContext.runAsNonRoot: true and runAsUser on steps. Never run build steps as root unless absolutely required.
  • Drop capabilities: set securityContext.capabilities.drop: ["ALL"] on step containers.
  • Do not use privileged: true in step security contexts.
  • Service account scoping: bind the minimum RBAC to the service account running the PipelineRun. Never use the pipeline service account with cluster-admin privileges.
  • Limit workspace access: use readOnly: true on workspace bindings when the task only needs to read.
yaml
# BAD -- running as root with no restrictions
steps:
  - name: build
    image: registry.redhat.io/ubi9/ubi-minimal
    script: |
      make build

# GOOD -- non-root with dropped capabilities
steps:
  - name: build
    image: registry.redhat.io/ubi9/ubi-minimal
    securityContext:
      runAsNonRoot: true
      runAsUser: 1001
      capabilities:
        drop: ["ALL"]
    script: |
      make build

Pipeline Structure Risks

  • Untrusted task references: taskRef with resolver: bundles or resolver: git can pull tasks from external sources. Pin to specific digests or commits.
  • Parameter injection: if pipeline parameters are passed into script: blocks via $(params.*), treat them as untrusted input. Validate and quote them.
  • When expressions: do not use when expressions as security controls. They skip tasks but do not enforce authorization.
  • Finally tasks: use finally: tasks for cleanup (removing credentials, revoking tokens) but never assume they will always run (node failure, timeout).
yaml
# BAD -- unpinned external task bundle
taskRef:
  resolver: bundles
  params:
    - name: bundle
      value: quay.io/someone/my-task:latest

# GOOD -- pinned by digest
taskRef:
  resolver: bundles
  params:
    - name: bundle
      value: quay.io/myorg/my-task@sha256:<digest>

Tekton Chains and Provenance

  • Enable Tekton Chains for automatic signing and attestation of TaskRun/PipelineRun results.
  • Verify signatures on built artifacts before promotion to production registries.
  • Store attestations alongside artifacts for auditability.

Containerfile / Dockerfile Hardening

Container image definitions are typically referenced or inlined in CI YAML. Misconfigurations here produce insecure images that run in production.

Run as Non-Root

  • Never leave USER unset -- the default is root. Always add a USER directive before the final CMD/ENTRYPOINT.
  • Create a dedicated user in the build stage.
dockerfile
# BAD -- runs as root by default
FROM registry.redhat.io/ubi9/ubi-minimal
COPY app /app
CMD ["/app/server"]

# GOOD -- explicit non-root user
FROM registry.redhat.io/ubi9/ubi-minimal
RUN microdnf install -y shadow-utils && \
    useradd -r -u 1001 appuser && \
    microdnf clean all
COPY --chown=appuser:appuser app /app
USER appuser
CMD ["/app/server"]

Secrets in Image Layers

  • Never COPY or ADD secrets (keys, credentials, .env files) into the image. Every layer is extractable.
  • Never use ARG or ENV for secrets -- they are visible in docker inspect and image history.
  • Use BuildKit secret mounts for build-time secrets. They are not persisted in layers.
dockerfile
# BAD -- secret baked into image layer
COPY .env /app/.env
ENV API_KEY=sk-secret-value

# GOOD -- BuildKit secret mount (not persisted in layer)
RUN --mount=type=secret,id=api_key \
    export API_KEY=$(cat /run/secrets/api_key) && \
    ./configure --with-key="$API_KEY"

Base Image Hygiene

  • Pin base images by digest, not mutable tags. :latest or :9 can change without notice.
  • Use trusted registries: prefer registry.redhat.io or your internal registry.
  • Use minimal base images: UBI minimal, micro, or distroless. Fewer packages = smaller attack surface.
  • Remove package managers from the final image when possible.
dockerfile
# BAD -- mutable tag, full OS image
FROM ubuntu:latest

# GOOD -- pinned digest, minimal trusted image
FROM registry.redhat.io/ubi9/ubi-minimal@sha256:<digest>

Multi-Stage Builds

  • Separate build and runtime stages. Build tools, compilers, and test dependencies must not ship in the final image.
  • Copy only the built artifact into the runtime stage.
dockerfile
# GOOD -- multi-stage: build tools stay in builder stage
FROM registry.redhat.io/ubi9/ubi:latest AS builder
RUN dnf install -y gcc make && dnf clean all
COPY . /src
RUN cd /src && make

FROM registry.redhat.io/ubi9/ubi-minimal@sha256:<digest>
COPY --from=builder /src/bin/app /app/server
USER 1001
CMD ["/app/server"]

Build Context Leakage

  • Always use .dockerignore (or .containerignore). Without it, the entire directory (including .git, .env, credentials, IDE configs) is sent to the build daemon.
  • Explicitly exclude sensitive paths.
text
# .dockerignore
.git
.env
*.pem
*.key
credentials/
node_modules/

Unnecessary Capabilities and Ports

  • Do not EXPOSE ports you don't use. While EXPOSE is documentation only, it signals intent and misconfigured orchestrators may open them.
  • Do not install sudo, su, or setuid binaries in the final image.
  • Set HEALTHCHECK so orchestrators can detect unhealthy containers.
dockerfile
# GOOD -- only expose what's needed, add healthcheck
EXPOSE 8080
HEALTHCHECK --interval=30s --timeout=3s \
  CMD curl -f http://localhost:8080/healthz || exit 1

Read-Only and Immutability

  • Design images to run with a read-only root filesystem at runtime. Write only to explicitly mounted volumes or tmpfs.
  • Avoid writing to the container filesystem in CMD/ENTRYPOINT scripts.

Common Containerfile Misconfigurations

MisconfigurationRiskFix
No USER directiveContainer runs as rootAdd USER <non-root> before CMD
COPY .env or COPY *.keySecrets baked into extractable layersUse BuildKit --mount=type=secret
ENV SECRET_KEY=valueSecret visible in docker inspectPass at runtime via orchestrator
FROM image:latestMutable tag; supply chain riskPin by @sha256:<digest>
No .dockerignore.git, credentials sent to build contextCreate .dockerignore with exclusions
Single-stage with build toolsCompilers, test deps ship to productionUse multi-stage builds
RUN chmod 777World-writable files in imageUse specific permissions (chmod 755)
No HEALTHCHECKOrchestrator cannot detect unhealthy stateAdd HEALTHCHECK instruction
sudo or setuid binaries installedPrivilege escalation pathRemove from final image

Common Misconfigurations (Both Platforms)

MisconfigurationRiskFix
Hardcoded secrets in YAMLCredential leakage via repo access or logsUse vault/CI variables/K8s Secrets
allow_failure: true on security jobsSecurity findings silently ignoredSet allow_failure: false
Mutable image tags (:latest)Supply chain compromise via tag poisoningPin by digest
Privileged containers/runnersContainer escape to hostUse unprivileged builds (Buildah/Kaniko)
Unscoped variables/secretsSecrets available to untrusted branches/jobsScope to protected branches/environments
Debug output (set -x, printenv)Secrets printed to job logsRemove debug flags; use masked variables
Unpinned remote includes/task refsExternal changes alter your pipelinePin to version/commit/digest
Missing security scan gatesVulnerable code reaches productionAdd SAST/SCA/image scanning as blocking steps
Auto-deploy to productionNo human review before deployRequire manual approval gate
Shell executor (GitLab)Job-to-job state leakage on shared hostUse Docker/Kubernetes executor
No USER in ContainerfileContainer runs as root in productionAdd non-root USER directive
Secrets in image layersCredentials extractable from imageUse BuildKit secret mounts
No .dockerignoreSensitive files leak into build contextCreate .dockerignore with exclusions
Single-stage buildBuild tools ship to productionUse multi-stage builds

Implementation Checklist

  • No secrets hardcoded in pipeline YAML; all credentials from vault or protected variables
  • Sensitive variables masked and scoped to protected branches/environments
  • Security scan jobs set to block pipeline on failure
  • All step/job images pinned by digest from trusted registries
  • Containers run as non-root with dropped capabilities; no privileged mode
  • Remote includes and task references pinned to specific versions/digests
  • Production deploys require manual approval
  • No debug output (set -x, printenv, env) in jobs that handle secrets
  • Pipeline parameters validated before use in scripts
  • Tekton Chains enabled for provenance and signing (where applicable)
  • Containerfiles set explicit non-root USER before CMD
  • No secrets in image layers, ARG, or ENV; use BuildKit secret mounts
  • .dockerignore / .containerignore excludes .git, .env, keys, credentials
  • Multi-stage builds separate build tools from runtime image
  • Base images from trusted registries, pinned by digest

Test Plan

  • Audit all .gitlab-ci.yml and Tekton Task/Pipeline YAML for patterns in the table above.
  • Verify that protected variables are not accessible from unprotected branches.
  • Confirm security scan jobs block merge on failure.
  • Test that pinned images resolve to the expected digest and are from trusted registries.
  • Review runner/service account RBAC for least privilege.

Frequently asked questions

What does the Build Yaml Misconfiguration AI skill do?

Apply when reviewing or writing GitLab CI (.gitlab-ci.yml), Tekton pipeline/task YAML, or Containerfiles/Dockerfiles used in CI builds. Detects common misconfigurations that expose secrets, weaken isolation, skip security gates, grant excessive privileges, or produce insecure container images.

Why use Build Yaml Misconfiguration on TypingMind?

Because you install it once and use it with any model. Build Yaml Misconfiguration is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Build Yaml Misconfiguration in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/RedHatProductSecurity/prodsec-skills/tree/main/module/skills/build-yaml-misconfiguration. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Build Yaml Misconfiguration?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Build Yaml Misconfiguration?

As many as you like. As long as a model supports skills, you can use Build Yaml Misconfiguration with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Build Yaml Misconfiguration AI skill free?

Yes. It is published on GitHub by RedHatProductSecurity under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇