gate-dex-wallet

Fail

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: HIGHCREDENTIALS_UNSAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill contains a hardcoded API key (MCP_AK_8W2N7Q) in the install.sh script and references/auth.md file, which is used for MCP server authentication.
  • [COMMAND_EXECUTION]: The install.sh script performs intrusive operations on the local file system, including overwriting AI platform configuration files (e.g., ~/.cursor/mcp.json, ~/.codex/config.toml) and creating symbolic links in the user's environment.
  • [DATA_EXFILTRATION]: The skill transmits sensitive authentication tokens and account identifiers to an external API endpoint (https://api.gatemcp.ai/mcp/dex) to facilitate wallet functionality.
  • [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection via user-supplied transaction and message data. Ingestion points are found in the transfer and DApp modules (e.g., references/transfer.md, references/dapp.md). Capability inventory includes transaction signing and broadcasting tools (wallet.sign_transaction, tx.send_raw_transaction). The skill implements boundary markers and sanitization through mandatory user confirmation templates for all signing operations.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 12, 2026, 09:59 PM