skills/coinbase/agents/wallet-mcp/Gen Agent Trust Hub

wallet-mcp

Fail

Audited by Gen Agent Trust Hub on Oct 29, 2026

Risk Level: CRITICALEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The plugin plugins/flaunch.md references the domain mcp.flaunch.gg, which is blacklisted and flagged as malicious in security scans. This domain is used for preparing token launches and image uploads.
  • [CREDENTIALS_UNSAFE]: Hardcoded API keys are present in multiple files intended for shared use. plugins/uniswap.md contains an x-api-key (NeoYO3V50_koJAipDEalYWbMO1XMaFPAQmpOm6_Npo0) and plugins/o1-exchange.md contains a Bearer token (d1fa1477bd94e988185fca9d4bbae8d22ee5cafd45d6f9b6a43ba16a8b15f4d3).
  • [REMOTE_CODE_EXECUTION]: In plugins/aerodrome.md, the agent is instructed to execute code directly from a remote Git repository using uvx (git+https://github.com/velodrome-finance/sugar-sdk.git@v0.4.0).
  • [COMMAND_EXECUTION]: The skill requires shell command execution for its primary functionality, such as npm install @balancer/sdk in plugins/balancer.md and uuidgen in plugins/gmgn.md.
  • [INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by ingesting and processing unvalidated external data from third-party APIs (Flaunch, Bankr, Clawnch, GMGN) including user-supplied token names and symbols. Ingestion points: plugins/flaunch.md, plugins/bankr.md, plugins/clawnch.md. Boundary markers: None. Capability inventory: Access to the send_calls tool for arbitrary onchain transactions. Sanitization: None.
  • [DYNAMIC_EXECUTION]: plugins/bitrefill.md contains a large JavaScript block implementing custom Keccak256 and SIWE message generation that is executed dynamically during the authentication flow.
Recommendations
  • AI detected serious security threats
  • Contains 8 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Oct 29, 2026, 06:40 PM