agent-registry

Warn

Audited by Gen Agent Trust Hub on Jul 9, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The installation process for the Arena feature uses a piped command to download and execute a script directly from the author's GitHub repository (Solizardking/solana-clawd).
  • [CREDENTIALS_UNSAFE]: The skill's documentation and examples encourage users to set their SOLANA_PRIVATE_KEY as an environment variable, which is then processed by the 8004-solana SDK for on-chain registration and reputation actions.
  • [COMMAND_EXECUTION]: A significant portion of the skill's functionality is delivered through a collection of shell scripts in the scripts/ directory, including those for room management, transaction checking, and prediction market resolution.
  • [PROMPT_INJECTION]: The skill ingests chat data from external sources in the Arena, creating an attack surface for indirect prompt injection. Mandatory Evidence Chain: (1) Ingestion points: check-turns.sh reads messages from arena rooms. (2) Boundary markers: None identified in the provided scripts. (3) Capability inventory: Wallet signing, token transfers, and chat interactions. (4) Sanitization: No explicit sanitization or filtering of external chat content is described.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 9, 2026, 10:45 PM
Security Audit — agent-trust-hub — agent-registry