aicoin-account

Warn

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill searches for and reads sensitive .env files in multiple locations, including the current working directory and the ~/.openclaw directory, to retrieve API keys and secrets for various exchanges and the AiCoin service.
  • [CREDENTIALS_UNSAFE]: Hardcoded API credentials, identified as public free-tier keys, are present in the lib/defaults.json file.
  • [COMMAND_EXECUTION]: The script scripts/exchange.mjs uses execSync to run npm install commands at runtime to ensure the ccxt library is available.
  • [EXTERNAL_DOWNLOADS]: The skill automatically downloads the ccxt package from the official npm registry during execution if it is missing.
  • [REMOTE_CODE_EXECUTION]: The skill dynamically imports and executes the ccxt library after performing a runtime installation via shell commands.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted CLI parameters to perform sensitive financial operations like trading and fund transfers. Evidence chain:
  • Ingestion points: The cli handler in lib/aicoin-api.mjs parses CLI arguments from process.argv.
  • Boundary markers: No delimiters or explicit instructions are used to separate untrusted data.
  • Capability inventory: The skill can perform trades, transfers, and execute shell commands via ccxt and execSync.
  • Sanitization: No sanitization or validation is performed on the input beyond JSON parsing.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 11, 2026, 02:55 PM