Skip to content

helm-guard

Static security analysis for Helm chart supply chain integrity

Get Started GitHub

Demo

helm-guard demo

How It Works

helm-guard uses a three-tier parser to analyze Helm charts without requiring the helm CLI. It checks Chart.yaml, values.yaml, and template files for supply chain risks that rendered-manifest scanners miss.

graph LR
    subgraph Input
        A["Chart.yaml"]
        B["values.yaml"]
        C["templates/*.yaml"]
    end

    subgraph "helm-guard"
        D["Tier 1: YAML Parser"] --> F[37 Security Checks]
        E["Tier 2: Regex Scanner"] --> F
        F --> G[FP Suppression]
    end

    subgraph Output
        H[JSON]
        I[SARIF 2.1.0]
        J[Text]
    end

    A --> D
    B --> D
    C --> E
    G --> H
    G --> I
    G --> J

Pipeline:

  1. Tier 1 (YAML) parses Chart.yaml, values.yaml, Chart.lock, values.schema.json with ruamel.yaml
  2. Tier 2 (Regex) scans template files as text for injection patterns (tpl, lookup, env)
  3. 37 Security Checks across 10 categories detect pinning, injection, trust, hooks, OLM, and CVE-based risks
  4. Tier 3 (Render) optionally runs helm template for rendered manifest checks (opt-in via --render)

Quick Example

$ helm-guard charts/my-chart --format text

Helm Chart Security Scan: charts/my-chart
Found 3 issue(s)

[CRITICAL] HLM-INJ-001: tpl function usage in templates
  File: templates/deployment.yaml:42
  Template uses tpl function enabling arbitrary template code execution
  Fix: Avoid tpl. Use direct value interpolation.

[HIGH] HLM-PIN-001: Chart dependency with SemVer range
  File: Chart.yaml:8
  Dependency 'redis' uses SemVer range '~1.2.0'. Pin to exact version.
  Fix: Pin to exact version (e.g., version: 1.2.3)

[HIGH] HLM-TRUST-002: Secrets in values.yaml
  File: values.yaml:15
  'database.password' has a non-empty default value
  Fix: Use empty defaults and set via --set or external secrets

Generate SARIF for GitHub Code Scanning:

helm-guard charts/my-chart --format sarif --output results.sarif

Comparison

Tool Chart-Level Template Injection Dependency Trust OLM Security
helm-guard ✅ ✅ ✅ ✅
Checkov ❌ ❌ ❌ ❌
Trivy ❌ ❌ ❌ ❌
Kubescape ❌ ❌ ❌ ❌
kube-linter ❌ ❌ ❌ ❌

helm-guard is the only tool that performs chart-level supply chain analysis. Existing tools scan rendered manifests for K8s best practices but miss dependency pinning, template injection, and chart provenance.

Features

  • Dependency Pinning

    Validates SemVer ranges, Chart.lock existence, image digest pinning, and OLM subscription channel versioning.

  • Template Injection Detection

    Detects tpl (CRITICAL), lookup, env/expandenv, getHostByName, and .Files.Get with user-controlled paths. All backed by real CVEs.

  • CVE-Based Checks

    Checks grounded in CVE-2020-11013, CVE-2023-25165, CVE-2024-25620, CVE-2025-53547, CVE-2026-35204, and CVE-2022-24348.

  • OLM Operator Security

    Validates OLM subscription auto-approval, catalog source trust, channel versioning, and operator namespace isolation.

What Gets Detected

37 checks across 10 categories:

Category Checks Examples
Pinning HLM-PIN-001..005 SemVer ranges, Chart.lock, image tags, OLM channels, SemVer compliance
Injection HLM-INJ-001..008 tpl, lookup, env, getHostByName, shell injection, .Files.Get, hardcoded registries
Trust HLM-TRUST-001..006 Missing schema, secrets, untrusted repos, hostNetwork, HTTP URLs, NetworkPolicy
Hooks HLM-HOOK-001..003 SecurityContext, delete policy, post-renderer scripts
OLM HLM-OLM-001..004 Auto-approval, community catalog, privileged namespace, unstable channels
Provenance HLM-PROV-001 Missing chart signature
Namespace HLM-NS-001..002 Privileged namespace, release namespace schema
Dependencies HLM-DEP-001..003 Security overrides, version conflicts, typosquatting
Security HLM-SEC-001..005 Path traversal, symlinked Chart.lock, valueFiles traversal, SA token automount

See Detection Rules for the full reference.

Next Steps