Back to skill
Skillv3.0.1
ClawScan security
Helm · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignMar 19, 2026, 9:54 AM
- Verdict
- Benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill is internally consistent with its stated purpose (a Helm chart helper) and runs a local helper script that invokes Helm; nothing indicates covert exfiltration or unrelated credential access.
- Guidance
- This skill appears to be a straightforward local Helm helper that runs scripts/script.sh to call your installed helm binary. Before installing: (1) confirm you have Helm v3+ installed (the script will refuse v2), (2) be aware it will create ~/.local/share/helm-helper and a log file there, (3) set KUBECONFIG / HELM_NAMESPACE as needed — the metadata does not declare these env vars but the script uses them, and (4) review the script if you want to confirm no additional network calls beyond normal Helm repo operations. If you don't want files written to your home, run the script manually or review/modify the DATA_DIR variables first. The author and homepage are listed (bytesagain.com); if you don't trust that source, inspect the full script before use.
Review Dimensions
- Purpose & Capability
- okName/description (Helm chart operations) align with the included script: the script implements create, lint, template, package, repo, list, status, etc. All required capabilities map to helm commands; curl is mentioned only in a suggested install instruction.
- Instruction Scope
- noteSKILL.md instructs the agent to run scripts/script.sh for each command which is expected. Minor inconsistency: SKILL.md lists a top-level 'KUBECONFIG' entry as if it were a command (shows `scripts/script.sh KUBECONFIG`) — in practice the script reads KUBECONFIG/HELM_NAMESPACE from environment and expects normal subcommands (create, lint, template, ...). The script logs locally and reads/writes chart directories, but does not attempt to transmit data to external endpoints other than normal Helm repo/network calls.
- Install Mechanism
- okThere is no install spec (instruction-only skill with an included shell script). No downloads or archive extraction are performed by the skill itself; the script only invokes locally installed helm and suggests how to install helm if missing.
- Credentials
- noteRegistry metadata declares no required env vars, but the script uses KUBECONFIG and HELM_NAMESPACE (and HELM_KUBECONFIG derives from KUBECONFIG). These are proportionate to the Helm use-case but should be documented in the skill metadata. No unrelated credentials or secrets are requested.
- Persistence & Privilege
- noteThe script creates a per-user data directory (~/.local/share/helm-helper) and a log file to store charts/packages/logs — reasonable for a CLI helper but it is persistent on disk. The skill is not always:true and does not modify other skills or global agent config.