Polygon Agent

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill relies on the polygon-agent CLI tool as its primary interface for executing on-chain actions and wallet management.
  • [EXTERNAL_DOWNLOADS]: The instructions require the user to install a global Node.js package (@polygonlabs/agent-cli) from the public NPM registry.
  • [REMOTE_CODE_EXECUTION]: The skill dynamically fetches additional instruction sets (sub-skills) from remote URLs hosted at https://agentconnect.polygon.technology/ to handle specific use cases like DeFi and Polymarket.
  • [CREDENTIALS_UNSAFE]: The skill manages highly sensitive data, including private keys and Sequence access keys. While it uses AES-256-GCM encryption for local storage (~/.polygon-agent/), it instructs users to provide private keys via command-line arguments (e.g., polygon-agent polymarket set-key <privateKey>), which may persist in shell command history.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing data from untrusted external sources.
  • Ingestion points: Data is fetched from x402-api.onrender.com in polygon-discovery/SKILL.md.
  • Boundary markers: Absent.
  • Capability inventory: Extensive on-chain capabilities (send, swap, pay) across all scripts.
  • Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 03:21 PM
Security Audit — agent-trust-hub — Polygon Agent