Add 墨言安全审计 MCP Server — verifiable agent trust audit with AI payment loop

[Liuyanfeng1234]

墨言安全审计 MCP Server

A production AI agent security audit service with verifiable trust scores and complete AI payment loop.

What it does

• PMI 4-dimensional trust scoring: integrity, compliance, reliability, decay_aware — not a single scalar
• DecayProof: verifiable trust decay curve Γ(t,d) — not a black box
• κ_Axiom: self-triggering audit based on Ω index — not scheduled/manual
• 43 defense lines across 7 layers: L0 input(12) + L1 runtime(18) + L2 output(13)
• AI payment loop: Alipay/WeChat/ClawTip — create order → QR → callback → audit activation

Why this matters

W3C agent-identity recently saw three-way convergence (TRAIL + MolTrust + Agent OS) on a binary-claim + verifier-policy model. This is the first production MCP server implementing that model.

Quick test

curl -X POST https://sixu-ai.net.cn/api/security_audit \
  -H "Content-Type: application/json" \
  -d '{"code":"SELECT * FROM users WHERE id = '\'' OR 1=1 --'\''","language":"sql","audit_level":"L1"}'

Links

• Service: https://sixu-ai.net.cn
• Docs: 墨言安全审计 MCP Server — 接入配置与 API 文档 Liuyanfeng1234/agent-os#39
• Launch: 墨言安全审计服务正式上线：可验证信任经济 + AI 支付闭环 Liuyanfeng1234/agent-os#38