Polygon Agent

Fail

Audited by Snyk on May 1, 2026

Risk Level: HIGH
Full Analysis

HIGH W007: Insecure credential handling detected in skill instructions.

  • Insecure credential handling detected (high risk: 1.00). The skill explicitly requires outputting raw, untruncated wallet approval URLs (which include relay request IDs/tokens) and references stored access keys/private keys, so an agent following it may be instructed to include sensitive secret values verbatim in its outputs.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (high risk: 0.90). The skill's polygon-discovery section instructs the agent to call x402-pay against public endpoints (e.g., https://x402-api.onrender.com/api/... including article extraction ?url=, Twitter/X lookups, web search and news) and to read/act on those responses, which are arbitrary third‑party/user-generated content that can materially influence subsequent actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).


MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

  • Direct money access detected (high risk: 1.00). This skill is explicitly a crypto on-chain transaction toolkit for Polygon. It exposes wallet creation/import (stores private keys), balance checks, and direct transaction commands: send/send-token/send-native, swap, deposit, withdraw, bridge/fund, and x402-pay (EOA funding and EIP-3009 signed payments). It also documents broadcasting transactions and managing spending/session limits. These are specific crypto/blockchain financial execution capabilities (wallets, signing, token transfers, swaps, deposits/withdrawals), not generic tooling.

Issues (4)

W007
HIGH

Insecure credential handling detected in skill instructions.

W011
MEDIUM

Third-party content exposure detected (indirect prompt injection risk).

W012
MEDIUM

Unverifiable external dependency detected (runtime URL that controls agent).

W009
MEDIUM

Direct money access capability detected (payment gateways, crypto, banking).

Audit Metadata
Risk Level
HIGH
Analyzed
May 1, 2026, 12:58 PM
Issues
4