Genlayer Claw Skill

• Outcome: Pass
• Latest audit: May 28, 2026, 4:04 AM
• Version:
  0.1.0

Overview

This is a documentation-only GenLayer explainer skill; its higher-risk blockchain examples should be treated as educational, not production-ready code.

Safe to install as educational reference material. Do not copy the LLM, web-fetch, escrow, treasury, staking, or dispute-resolution examples into production without independent security review, validated data sources, privacy controls, bounded model outputs, human or governance approval for valuable actions, and clear rollback or appeal paths.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (5)

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The guide shows fetching external web data inside a contract but does not warn that request contents and derived parameters may be transmitted off-platform to third-party services. In a smart-contract context, developers may incorrectly assume contract inputs are confined to the chain, leading to unintended disclosure of sensitive data, metadata, or user-linked activity.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The LLM example sends user-provided text to an AI processing function without clearly warning that prompts may be transmitted to an external model provider or otherwise leave the deterministic on-chain boundary. In this context, developers could pass confidential contract inputs, personal data, or adversarial text into a model call without understanding the privacy and prompt-injection risks.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The example shows a public write function fetching external proof data and automatically transferring escrowed funds based on an AI evaluation, but it does not warn that untrusted remote content influences a financial action. In this context, readers may copy the pattern into production without adding consent, validation, or human review, creating a real risk of unsafe fund release from manipulated or unavailable web data.

Missing User Warnings

Medium

• Confidence: 88% confidence
• Finding: This example demonstrates autonomous retrieval of market data and immediate treasury reallocation based on LLM output, yet it omits any warning about the financial and governance consequences of that automation. Because treasury management is highly sensitive, the undocumented pattern encourages unsafe designs where manipulated external data or unstable model behavior can cause asset misallocation.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The dispute-resolution example uses AI output to determine a winner and distribute funds automatically, but it does not warn that subjective model output is directly controlling asset movement. In a dispute context, ambiguous evidence, prompt manipulation, or model inconsistency could produce unfair or irreversible payouts if developers follow the example as written.

VirusTotal

64/64 vendors flagged this skill as clean.