The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The tool setup script (scripts/setup_tools.sh) includes documentation strings recommending the installation of the Helm CLI using a script from the official project repository. Analysis of the source confirms these are static instructions printed to the console and do not involve automated execution of remote code by the skill itself.
[EXTERNAL_DOWNLOADS]: The skill documentation (test/test-crd-chart/README.md) includes references to download and apply YAML manifests from the official cert-manager project repository to assist users in configuring cluster dependencies.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external, untrusted Helm charts, which creates a potential surface for indirect instruction injection via manifest content.
Ingestion points: Helm chart directory contents, specifically Chart.yaml, values.yaml, and manifest files within the templates/ directory.
Boundary markers: Not present; the skill processes the user-provided chart directory directly as the input for its validation stages.
Capability inventory: The skill executes local shell scripts (validate_chart_structure.sh, generate_helpers.sh) and invokes system CLI tools including helm, yamllint, and kubeconform to perform validation tasks.
Sanitization: The skill's internal Python parsing logic uses yaml.safe_load to process resource metadata, mitigating standard YAML-based injection vectors.

helm-validator