gator-cli

Fail

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: HIGHCREDENTIALS_UNSAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill manages Ethereum private keys and explicitly documents that they are stored in plaintext JSON format at ~/.gator-cli/permissions.json. It also provides instructions for configuring sensitive credentials like API keys and IDs for delegation storage.
  • [COMMAND_EXECUTION]: The skill uses the gator CLI to perform high-privilege blockchain operations, including wallet initialization (init), upgrading accounts to smart accounts (create), and executing transactions on-chain (redeem, revoke).
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of the @metamask/gator-cli package from the official NPM registry to function.
  • [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection by processing external data, such as Solidity function signatures and raw call data, which are then signed and executed as transactions.
  • Ingestion points: Command line arguments for parameters like --function, --selectors, --args, and --callData in SKILL.md.
  • Boundary markers: None provided in the skill instructions to separate untrusted data from the command context.
  • Capability inventory: Signing and broadcasting transactions via the gator CLI across multiple EVM-compatible chains.
  • Sanitization: No explicit sanitization or validation of input data is mentioned before it is passed to the CLI for execution.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 29, 2026, 07:36 AM
Security Audit — agent-trust-hub — gator-cli