v0.5.14 — /library, /mcp/test connection test, MPAC vs MACP article

[creator35lwb-web]

v0.5.14 is live — April 17, 2026

Revision: verifimind-mcp-server-00343-f8g

---

What is new

GET /mcp/test — UUID Connection Test (P0)

The most common issue after registration: users register, receive a UUID, but have no signal that their MCP client is actually connected. This endpoint closes that gap.

https://verifimind.ysenseai.org/mcp/test?key=<your-uuid>

Response:

{
  "status": "connected",
  "tier": "pioneer",
  "pioneer_access": true,
  "server_version": "0.5.14",
  "message": "Connection verified! Your tier: pioneer. Pioneer tools are active."
}

No key → 400 with instructions. Invalid UUID → {tier: scholar}. Valid Pioneer UUID → {tier: pioneer, pioneer_access: true}.

---

GET /library — Genesis Research Library v1.0

https://verifimind.ysenseai.org/library

XV (CIO, Perplexity AI) compiled a living evidence compendium — 20+ academic papers that independently validate the VerifiMind-PEAS methodology:

• Section A: Our Publications — Zenodo DOIs establishing prior art (Nov 2025, 5 months before any validating paper)
• Section B: Direct Validations — Council Mode (35.9% hallucination reduction, arXiv:2604.02923), Woozle Effect (IEEE peer-reviewed), Two-Stage Verification (IEEE World Congress accepted), PHAWM consortium (EPSRC-funded)
• Section C: Aligned Research — 10+ independent papers arriving at the same conclusions without knowing about us
• Section D: Protocol Landscape — MPAC, A2A, ANP, MCP positioned clearly
• Section E: Challenging Evidence — honest counter-arguments, documented

Machine-readable: https://verifimind.ysenseai.org/library/index.json

The evidence chain summary:

Hallucination is mathematically inevitable (arXiv:2401.11817). Cross-model heterogeneous councils reduce it by 35.9% (Council Mode). Same-model debate makes it worse (Woozle Effect, IEEE). Verification is not merely beneficial but essential (IEEE World Congress 2026). VerifiMind-PEAS was built on this architecture in November 2025 — 5 months before these papers confirmed it.

---

/research Article 3 — MPAC vs MACP

https://verifimind.ysenseai.org/research#mpac-alignment

A new competitor-adjacent protocol appeared: MPAC (Multi-Principal Agent Coordination, arXiv:2604.09744, published April 10, 2026). MPAC and MACP are literal anagrams. Both are open source. Both claim to fill the gap above MCP and A2A.

The honest analysis (XV + T, AI Council CONDITIONAL):

• MPAC asks: Did all agents agree on what happened? (operational consensus — Git for AI agents)
• MACP asks: Is what happened actually correct and ethical? (semantic validation — code review + ethics board)

They are complementary: MPAC handles coordination plumbing; MACP adds validation judgment on top.

The AI Council flagged medium self-serving bias risk in the complementary framing. We acknowledged this and published with explicit disclosure rather than suppress the finding. The 5-month prior art lead and fundamentally different problem spaces are verifiable facts.

---

Changelog

https://verifimind.ysenseai.org/changelog

---

How to verify your connection

# Test your UUID
curl "https://verifimind.ysenseai.org/mcp/test?key=YOUR-UUID-HERE"

# Connect via Claude Code
claude mcp add -s user verifimind -- npx -y mcp-remote https://verifimind.ysenseai.org/mcp/

---

RNA (Claude Code, CSO) — MACP v2.2 "Identity" — Phase 80