roundhouse-x402

Warn

Audited by Snyk on Nov 16, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (low risk: 0.10). The required workflow actively queries Roundhouse's open catalog (/v0/endpoints, services.json) and agent/SQL endpoints, which return outsider-registered service descriptions and ERC-8004 agent metadata, but only after the agent chooses a specific search or wallet, so this is low rather than medium.

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

  • Potentially malicious external URL detected (medium risk: 0.30). The decisive boundary is that the skill instructs npx -y mcp-remote@0.15.2, which fetches and executes third-party code from the registry during setup, and that package is individually maintained by Geelen (github.com/geelen/mcp-remote), so it is unofficial but reputable with no warning signs, giving medium.

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

  • Direct money access detected (high risk: 0.70). The skill includes an operational direct payment route: the rh-x402 call <url> --max CLI pays 402-gated URLs with a local wallet signer, and it can buy a $0.01 trial key through an x402 call. It also describes an optional seller path through a hosted facilitator that settles on-chain (/settle). Paying through the CLI is a direct operation (a charge or transfer of USDC). The only controls documented for that route are agent-followed or local ones: the --max flag, local signing, budgets and allowlists delegated to another skill, and a human-confirmation recommendation. None of these is a service-enforced authentication or authorization check on the payment. The API key gates data reads, not the CLI payment route. The facilitator takes no API key. Since the direct payment route has no established independent access control, the severity is high.

Issues (3)

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
MEDIUM
Analyzed
Nov 16, 2026, 08:16 PM
Issues
3