Security audit

Abstract Toolkit

Security checks across malware telemetry and agentic risk

Overview

This Abstract blockchain toolkit is coherent, but it gives an agent raw private-key authority to move real mainnet funds and call arbitrary contracts without built-in confirmation safeguards.

Install only if you intentionally want an agent-accessible Abstract mainnet toolkit. Use a dedicated low-balance wallet, prefer testnet first, avoid long-lived raw private keys, and manually verify every chain, address, router, ABI, token, amount, approval, and transaction before allowing any write operation.

SkillSpector

By NVIDIA

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (6)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 93% confidence
Finding: The skill instructs users to export and use a private key via environment variables, but the skill declares no permissions despite requiring access to sensitive credentials. This creates a trust and transparency gap: an agent could access signing material and perform blockchain transactions without an explicit permission boundary, increasing the risk of unintended fund movement or key exposure in logs, subprocesses, or inherited environments.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill provides direct instructions for irreversible mainnet actions such as bridging, deploying contracts, transferring assets, swapping tokens, and arbitrary write calls, but it does not prominently warn that these actions spend real funds and cannot be undone. In an agent context, this is dangerous because automated execution may cause immediate financial loss, deployment mistakes, or malicious/incorrect contract interactions on a live chain.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The documentation instructs users to export a raw private key into an environment variable and immediately use it in a script, but it provides no warning about the sensitivity of the credential or safer handling practices. In an agent skill focused on blockchain deployment and asset movement, this increases the chance that operators expose high-value signing keys through shell history, logs, shared environments, or reuse of funded EOAs, which can lead to wallet compromise and asset theft.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The script sends approval and swap transactions immediately once invoked, with no interactive confirmation, dry-run mode, or secondary verification of token addresses, router address, amounts, or quoted output before broadcasting irreversible on-chain actions. In an agent skill context, this is more dangerous because automated or indirect invocation can cause unintended asset movement, token approvals, and slippage exposure without explicit human review at execution time.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The script sends ETH immediately after parsing CLI arguments, with no confirmation prompt, no dry-run mode, and no explicit warning that the transfer is irreversible. In an agent skill that automates on-chain actions, this raises the risk of accidental or manipulated fund transfers caused by bad parameters, prompt injection into the calling workflow, or operator error.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The token transfer path invokes ERC-20 transfer directly with no confirmation or irreversible-action warning. This is particularly risky in this skill context because the script accepts arbitrary token addresses when the symbol is unknown, so an agent or user can be led into sending valuable tokens to the wrong address without any final review step.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.