Back to skill
Skillv1.0.0
ClawScan security
Ads Creative Scoring · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignMar 3, 2026, 3:51 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- This is an instruction-only ad-creative scoring skill that asks for no credentials, installs, or file access and its runtime instructions are coherent with its stated purpose.
- Guidance
- This skill is instruction-only and internally consistent with its stated purpose. Before installing: 1) confirm your agent/platform does not automatically transmit the 'handoff payload' to external endpoints (the SKILL.md suggests producing such a payload but does not specify a destination); 2) treat outputs as advisory — use the skill alongside human review and official platform policy checks (this skill explicitly warns not to fabricate policy outcomes); 3) test with dummy or non-sensitive creative data first to validate behavior; 4) if you plan to wire the handoff into automation (e.g., ticketing or publishing), explicitly control and audit the destination and required credentials. No high-risk indicators were found in the provided files.
Review Dimensions
- Purpose & Capability
- okThe name/description (ads creative scoring across multiple platforms) matches the SKILL.md content: scoring rubric, QA notes, test plans, and platform-specific guidance. There are no unexpected required binaries, env vars, or config paths that would be disproportionate to the stated goal.
- Instruction Scope
- okSKILL.md contains concrete instructions for generating objectives, variants, QA checks, and decision rules. It does not instruct the agent to read arbitrary local files, exfiltrate data, call unknown external endpoints, or access credentials. One note: the 'escalate with a structured handoff payload' instruction implies preparing data for another system but does not define any destination or transmission mechanism; that is a procedural detail rather than a hidden behavior.
- Install Mechanism
- okNo install spec and no code files — the skill is instruction-only. This minimizes filesystem and network installation risk.
- Credentials
- okThe skill declares no required environment variables, credentials, or config paths. Nothing in the SKILL.md attempts to read undeclared secrets or external keys.
- Persistence & Privilege
- okFlags indicate default behavior (not always:true) and autonomous invocation is allowed (platform default). The skill does not request persistent system presence or modify other skills/configuration.