simmer-x402

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt instructs users/agents to place private keys inline (export EVM_PRIVATE_KEY=0x...) and shows API keys passed directly in command headers (Authorization: Bearer sk_live_...), which requires embedding secrets verbatim in generated commands/outputs and is high-risk.

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

• Third-party content exposure detected (high risk: 0.85). The skill (x402_fetch in x402_cli.py and the SKILL.md fetch examples) explicitly fetches arbitrary public URLs (e.g., api.kaito.ai, quicknode/x402, CoinGecko or any provided ), parses 402 payment bodies and returned JSON, and uses that data to decide payments and return content — meaning untrusted third‑party responses can materially influence payment decisions and subsequent agent behavior.

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

• Direct money access detected (high risk: 1.00). The skill is explicitly designed to perform on-chain payments: it autonomously makes x402 payments using USDC on the Base network, requires a wallet private key (EVM_PRIVATE_KEY/WALLET_PRIVATE_KEY), signs USDC transfer authorizations, and retries requests with a payment signature. It provides CLI commands and a Python API (x402_fetch) that automatically pay 402-gated endpoints and can check wallet balances and RPCs. These are specific crypto payment/transaction capabilities (wallet signing and USDC transfers), not generic tooling.