Agent Fuel

Plain-language checklist before installing or running this skill: - What it will do: the skill uses the MoonPay CLI ('mp') and your locally-authenticated MoonPay account to check balances, buy/swap tokens and send funds (including automatically paying x402 payment requests). If configured to run as a daemon, it will do these actions periodically. - Major red flags to address before use: - The code can pay arbitrary addresses derived from x402 headers. Make sure you understand and trust the services that will receive payments, or implement a recipient whitelist. The README mentions 'whitelist mode' and a 'kill switch' but the provided code does not enforce either. - The shell script uses eval on a configurable NOTIFY_CMD. If an attacker or misconfigured file can set that field, arbitrary commands could run. Treat NOTIFY_CMD as unsafe unless you control it and avoid embedding user-supplied content there. - The TypeScript and shell code build shell command strings (execSync/`mp ...` and eval) using config values. This creates command-injection risks if config or payment headers can be manipulated. - The skill does not declare the sensitive credentials it will use (MoonPay auth is expected but not declared). The skill will rely on your local 'mp login' state to access funds. - Mitigations and recommendations: - Audit the code locally. If you want to use it, run it in a safe environment (test account / low-funding wallet) first. - Add or require an explicit whitelist of allowed payment recipients and/or require human confirmation for top-ups/payments above a small threshold. - Replace eval usage: change NOTIFY_CMD handling to a safer invocation (e.g., spawn with args array) and avoid evaluating arbitrary strings. - Sanitize and avoid interpolating user-provided strings into shell commands. Use child_process spawn with argument arrays or validated tokens/currency names. - Keep maxDailySpend and x402MaxPerRequest very low during testing and ensure logging/alerts are routed to a channel you control. - Consider running the skill with a dedicated MoonPay account or test wallet that only holds limited funds. - Final decision guidance: if you need autonomous on-chain top-ups and are comfortable reviewing and hardening the code (implement whitelist/kill-switch, remove eval, sanitize inputs), this skill is usable. If you cannot review and secure it, do not grant it access to a funded MoonPay account — treat it as suspicious and test only with minimal funds.

These patterns may indicate risky behavior. Check the VirusTotal and OpenClaw results above for context-aware analysis before installing.