skills/helius-labs/core-ai/phantom/Gen Agent Trust Hub

phantom

Fail

Audited by Gen Agent Trust Hub on Apr 19, 2026

Risk Level: CRITICALEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill recommends installing the author's own MCP server via npx helius-mcp@latest and utilizes several official Solana and Phantom SDKs from NPM.
  • [COMMAND_EXECUTION]: Instructs users to configure the environment using CLI commands like claude mcp add helius npx helius-mcp@latest and helius signup.
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface (Category 8).
  • Ingestion points: The skill processes untrusted data from the Solana blockchain, including NFT metadata and transaction descriptions, via Helius DAS and Enhanced Transactions APIs (referenced in references/helius-das.md and references/helius-enhanced-transactions.md).
  • Boundary markers: Absent. Code examples in references/integration-patterns.md render external data directly without explicit boundary delimiters or warnings.
  • Capability inventory: The skill facilitates signing messages and submitting transactions to the blockchain via Helius Sender (references/helius-sender.md).
  • Sanitization: Sanitization logic is not explicitly defined in the provided React and Node.js code snippets.
  • [EXTERNAL_DOWNLOADS]: An automated scanner flagged https://sandbox.phantom.dev as a phishing URL. This is a legitimate developer resource provided by the Phantom wallet team and is likely a false positive.
  • [COMMAND_EXECUTION]: An AV scanner flagged SKILL.md with HttpRequest-inf. This is likely a false positive caused by the extensive documentation of API endpoints and HTTP request examples provided for developer guidance.
Recommendations
  • CRITICAL: 1 infected file(s) detected - DO NOT USE
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Apr 19, 2026, 07:10 PM