Spraay Compute & Futures

• Outcome: Review
• Latest audit: Jun 2, 2026, 1:25 AM
• Version:
  1.0.0

Overview

This skill is openly about paid USDC compute, but its broad auto-trigger guidance could cause agents to spend wallet funds on generic AI requests without clear user confirmation.

Review before installing if your agent has access to a funded wallet. Use spending caps, require explicit approval before any paid request or compute-futures deposit, prefer free estimate/model endpoints first, and avoid enabling this skill as the default handler for generic inference or media-generation requests.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The README explicitly promotes wallet-based USDC payments, deposits, prepaid balances, and refunds on Base and Solana mainnet, but it does not clearly warn users that these are real-money blockchain transactions affecting live assets. In an agent-skill context, that omission is security-relevant because autonomous or semi-autonomous agents may trigger deposits or compute-futures funding without the operator fully appreciating that funds can be committed, spent, or exposed to irreversible on-chain actions.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The trigger guidance is very broad and encourages invocation for generic phrases like 'run inference' or 'generate an image/video,' even when the user did not explicitly request this provider or realize it performs paid actions. In a wallet-connected agent, this can cause unintended activation of a spending-capable skill and lead to unnecessary or unauthorized USDC charges.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The skill prominently describes paid compute and prepaid deposits but does not clearly warn that actions spend real USDC from a funded on-chain wallet, including potentially larger prepaid deposits for compute futures. In the context of an autonomous or semi-autonomous agent, missing spend disclosure materially increases the risk of surprise financial loss because users may interpret these as ordinary API calls rather than blockchain payments.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The quickstart demonstrates wallet-backed x402 payment flows and prepaid deposits/refunds using real USDC, but it does not place a prominent warning up front that examples may trigger real on-chain spending. In an agent skill context, users may treat quickstarts as safe copy-paste snippets, so omission of explicit spend confirmation increases the risk of unintended charges or deposits.

VirusTotal

64/64 vendors flagged this skill as clean.