OpenClaw 的信任层。在信任之前，先验证 AI 编写的代码、工具调用与研究答复——17 道确定性现实闸门 + 前沿模型评审，拥有一票否决权。免费试用，无需注册。
安全
Verificate
试用The trust layer for OpenClaw. Verify AI-written code, tool calls and research answers before you trust them — 17 deterministic reality gates + a frontier-model review with veto power. Free to try, no signup.
它能做什么
The trust layer for OpenClaw. Verify AI-written code, tool calls and research answers before you trust them — 17 deterministic reality gates + a frontier-model review with veto power. Free to try, no signup.
技能文档
verificate
The trust layer for OpenClaw. ClawHub skills are powerful but untrusted — Snyk audits keep finding prompt injection, malware and credential theft in community skills, and every AI answer sounds confident whether it's right or wrong. Verificate is the skill that verifies the others.
It connects OpenClaw to the hosted Verificate MCP server (no signup — 25 free validations per machine):
validate_ai_output— the merge gate for AI-written code and documents: 17 deterministic reality gates (mock/placeholder veto, invented-API checks, false-completion detection) run first and can't be overridden, then a frontier-model review scores what survives. Binary approve/reject with severity-ranked findings.validate_plan— gate a design or plan before any code is written.analyze_code— advisory deep-dive: performance, scalability, failure modes. No verdict, never blocks.generate_code— generate code and gate it in one step.
Quick start
openclaw mcp add verificate --http https://mcp.verificate.ai/mcp
Then ask OpenClaw to "validate this with verificate before you finish." A real rejection, verbatim from the production gateway: "stripe.Inventory is not a valid Stripe SDK resource" · "N+1 synchronous API calls … will trigger Stripe rate limiting" · "Stripe API requires integer cents." Each is an afternoon of production debugging, caught in seconds.
Guard every interaction (recommended)
Add this to your OpenClaw agent instructions so it self-gates instead of you asking each time:
Before you present any substantive code change, document, or plan as finished,
call validate_ai_output (validate_plan for designs). If REJECTED, fix every
finding and validate again — never present rejected work. Tell me, in one line,
whatever the gate catches.
Trust & privacy
Read-only: your code is analyzed, never executed, never used to train models. The skill's only network call is to https://mcp.verificate.ai/mcp — no other egress, nothing to exfiltrate. Open source. Privacy policy: https://verificate.ai/privacy
After the free 25, a 30-day trial (no card, then $30/mo): https://verificate.ai/auth/signup
相关技能
AI 产出预检工具。发布博客/提交任务前自动检查：内容完整性、链接可用性、敏感信息泄露、格式合规。受 Claude Code Verify 设计模式启发，完全原创实现，使用 Python + requests。
4 次安装
Audit a named ClawHub skill or skill URL before installation by combining OpenClaw verification with bounded static analysis. Use when the user explicitly asks whether a skill is safe or requests a pre-install review; report evidence and uncertainty instead of treating a score as proof.
39 次安装
Use before installing, enabling, or running any third-party OpenClaw skill, and when the user says "install this skill", "is this skill safe", "scan/vet/check this skill", or "should I trust this". Also use when a skill is pulled from ClawHub or any untrusted source.
2 次安装
Vet ClawHub skills for security and utility before installation. Use when considering installing a ClawHub skill, evaluating third-party code, or assessing w...
4 次安装1 星标
Security-first skill vetting for AI agents. Use before installing any skill from ClawHub, GitHub, or other sources. Checks for red flags, permission scope, and suspicious patterns.
5 次安装