Skillv1.0.0

ClawScan security

Image Detection · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

SuspiciousFeb 21, 2026, 5:35 PM

Verdict: suspicious
Confidence: medium
Model: gpt-5-mini
Summary: The skill's instructions describe local and cloud image-detection features but the registry metadata omits required credentials and install steps, and it references third-party npm packages and external APIs (which could expose images) without declaring or justifying them.
Guidance: Before installing, verify the following: 1) Ask the publisher for a source repo or homepage and versioned releases—avoid running npm install on unknown scoped packages without reviewing their code. 2) Confirm how Hive Moderation or other cloud APIs are authenticated (the skill should declare required env vars like HIVE_API_KEY); never paste production credentials into an untrusted skill. 3) Be cautious with reverse image search or cloud-based moderation—these will upload images and could leak sensitive content. 4) If you need privacy, prefer the 'Privacy-Focused Stack' but still inspect the local npm packages for telemetry or build-time scripts. 5) Run installs in a sandbox, pin package versions, and review the packages' maintainers and license. If the publisher cannot provide a repository, clear credential-handling guidance, and package provenance, treat this skill as higher risk.
Findings: [no-code-files-or-scan-target] expected: The static regex scanner found no code files because this is an instruction-only skill (SKILL.md). Absence of scan findings does not imply safety; the instructions themselves reference npm packages and external APIs that the scanner could not evaluate.

Review Dimensions

Purpose & Capability: noteThe SKILL.md describes both local analysis and a cloud-based Hive moderation API, which matches the stated purpose. However the registry metadata claims no install steps, no required env vars, and no primary credential, which is inconsistent with a skill that integrates with external APIs and third-party npm packages.
Instruction Scope: concernRuntime instructions tell the agent to npm install several scoped packages and to integrate with the Hive Moderation REST API and reverse-image-search (which may upload images). The SKILL.md does not specify how API credentials should be provided or protected, nor does it limit what images may be sent to external services—this can lead to unintended data exfiltration.
Install Mechanism: concernThere is no formal install spec in the registry, but SKILL.md instructs running npm install for scoped packages (@clawhub/* and @raghulpasupathi/*). Installing arbitrary npm packages at runtime is moderate risk: packages can execute code during install and are from unknown publishers. No pinned versions, checksums, or well-known release hosts are provided.
Credentials: concernThe metadata declares no required environment variables or primary credential, yet the Hive API and some model hubs typically require API keys/tokens. This omission is a red flag—credentials are necessary for cloud services but are not declared or explained.
Persistence & Privilege: okThe skill does not request always: true and has no declared system-wide persistence or privileges. It appears to be user-invocable and not forced into every agent run.