monad-swarm-agent

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
  • [CREDENTIALS_UNSAFE] (MEDIUM): The skill requests a MONAD_PRIVATE_KEY environment variable for blockchain interactions. While used locally via the trusted ethers library, handling raw private keys is a high-sensitivity operation.
  • [COMMAND_EXECUTION] (MEDIUM): The skill relies on local execution of TypeScript scripts (log-to-monad.ts, swarm-analyze.ts) using ts-node. Although the logic is straightforward simulation and signing, local script execution requires trust in the author.
  • [EXTERNAL_DOWNLOADS] (LOW): The scripts use npx ts-node, which may download the runner at runtime if not present. It also depends on the ethers library.
  • [PROMPT_INJECTION] (SAFE): No malicious prompt injection patterns were detected in the instructions or metadata.
  • [DATA_EXFILTRATION] (SAFE): No evidence of sensitive data being sent to unauthorized external domains was found. Network activity is limited to the configured Monad RPC URL.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:30 PM