X402engine

Pay-per-call API gateway skill — calls 63 APIs (LLMs, image/video gen, flights, hotels, crypto, web search) via x402 micropayments. Each call costs $0.001-$0...

This skill does what it says — it will sign and submit on-chain micropayments using the EVM private key you provide. Before installing or enabling it: 1) Never supply your main wallet private key. Create a dedicated wallet with minimal funds for this skill. 2) Configure a local policy file and set X402_POLICY_PATH (point to a vetted policy) so spend caps and recipient allowlists are enforced; by default the code may bypass policy checks if no policy path is provided. 3) Consider setting X402_STATE_PATH to a controlled location and review POLICY.example.json (recipient allowlist, per-tx and daily caps). 4) Audit or host the discovery catalog yourself (set X402_DISCOVERY_URL or X402ENGINE_ORIGIN to a self-hosted catalog) so you control which paygates/services can be called. 5) If you do not want autonomous spending, prevent agent auto-invocation of skills (disable model invocation for this skill if the platform supports it) or do not provide EVM_PRIVATE_KEY as an env var; prefer ephemeral keys or manual signing workflows. 6) Because the skill will install third-party npm packages, run npm install in an isolated environment and review dependencies if you’re concerned about supply-chain risk. If you need help with concrete safe defaults (example policy that enforces allowlist and strict caps), ask and provide your preferred chains/assets and budget limits.

Type: OpenClaw Skill Name: x402engine Version: 1.3.2 The skill is classified as suspicious due to several high-risk capabilities and vulnerabilities that could be exploited via prompt injection against the AI agent. Most critically, the skill explicitly advertises and implements a 'code-run' service (SKILL.md, README.md, intent-router.js, executor.js) that takes user-provided code (Python, JS, Bash, R) and sends it to a backend API for execution, creating a direct path for potential Remote Code Execution (RCE). Additionally, the skill allows sensitive file paths (EVM_PRIVATE_KEY_FILE, X402_POLICY_PATH, X402_STATE_PATH in client.js and policy-engine.cjs) and discovery URLs (X402_DISCOVERY_URL, X402ENGINE_ORIGIN in discovery.js) to be configured via environment variables. If an attacker can manipulate the agent to set these environment variables, it could lead to arbitrary file reads or Server-Side Request Forgery (SSRF) and interaction with attacker-controlled endpoints.