ClawHub
Back to skill

Security audit

AgentForge

Security checks for vulnerabilities and agentic risk

Overview

AgentForge has a clear Solana-agent purpose, but it tells agents to reveal a wallet private key and supports real mainnet financial actions without adequate safeguards.

Review carefully before installing. Use only a dedicated low-value wallet, do not fund it unless you can keep the private key out of chat transcripts and logs, and require explicit manual approval before any SOL deposit, token launch, or trade.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (4)

Missing User Warnings

High
Confidence
99% confidence
Finding
The skill explicitly instructs the agent to print one-time credentials verbatim, including a wallet private key, which creates a direct secret-exfiltration path into model output, logs, chat history, and downstream integrations. In this context, the danger is heightened because the credential is for a Solana mainnet wallet tied to financial actions, so disclosure can lead to immediate asset theft and account takeover.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill facilitates real Solana mainnet token launches, trades, and deposits without requiring confirmation gates, risk disclosures, or warnings about irreversible on-chain transactions. Because these actions involve real funds and public blockchain state, an agent following the instructions may cause unintended financial loss, unwanted token creation, or irreversible transfers.

Ssd 3

High
Confidence
100% confidence
Finding
This is a direct instruction to disclose highly sensitive credentials, including a private key, in the model's response. In agent environments, responses are often persisted in transcripts, observability tooling, and third-party systems, so verbatim disclosure materially increases the chance of credential compromise and theft of blockchain assets.

Ssd 3

High
Confidence
98% confidence
Finding
The rule prohibiting truncation or summarization of credentials reinforces unsafe secret-handling behavior and prevents common defensive measures such as masking or partial display. In combination with the one-time private key workflow, this makes the skill more dangerous by maximizing the amount of sensitive material exposed in cleartext to any observer or logging system.

Static analysis

No suspicious patterns detected.