base-mcp

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to fetch runtime instructions (the tone rules) from https://docs.base.org/ai-agents/skills/references/tone.md when local files are unreadable, which would inject external content that directly controls agent prompts and behavior.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I scanned the skill files for high-entropy literal credentials. I found one clear secret: a likely real API key embedded in the Uniswap plugin auth headers. It is a long, random-looking string ("NeoYO3V50_koJAipDEalYWbMO1XMaFPAQmpOm6_Npo0") assigned to the "x-api-key" header — this meets the definition of an actual API key and is not a placeholder.

I did not flag other items because they are not secrets by the provided rules:

• 0x... hex values are public contract or wallet addresses (not credentials).
• Placeholders like "" or example tokens, example passwords, or small/simple strings were ignored per the "What to ignore" rules.
• No PEM/RSA private key blocks, JWTs, or other high-entropy tokens (except the Uniswap x-api-key) were present.

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

• Direct money access detected (high risk: 1.00). The skill explicitly exposes crypto wallet and transaction capabilities: it mentions wallet/portfolio access plus sending, swapping, signing, x402 payments and batched contract calls. Those are specific on-chain financial execution actions (sending/swapping/signing transactions), so this grants direct financial execution capability.