gmgn-wallet-score

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes system network utilities (ifconfig or ip addr show) to enumerate local network interfaces and detect IPv6 addresses for troubleshooting. It also executes a third-party CLI tool (gmgn-cli) using subprocess.run in a Python script.
  • [DATA_EXFILTRATION]: It performs a network request to an external service (ipv6.icanhazip.com) to determine the host's public IP address.
  • [CREDENTIALS_UNSAFE]: The instructions direct the agent to run gmgn-cli config and display the output to the user if the initial configuration check fails; this carries a risk of exposing the stored API key if the CLI tool does not mask it in the output.
  • [INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect injection as it processes untrusted data from the GMGN API. 1. Ingestion points: Wallet stats, transaction activity, and token metadata in SKILL.md via CLI calls. 2. Boundary markers: No delimiters or ignore-instructions are used when reflecting external strings (e.g., ENS names, Twitter handles) into the response. 3. Capability inventory: The skill can execute shell commands and generate Python code. 4. Sanitization: While numeric data is cast to floats, text data from the API is used without sanitization or escaping.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 02:48 AM