X-Z-CS AI validation with auditable reasoning. 8 active tools; 5 in maintenance. v0.5.64
VerifiMind Genesis MCP Server (io.github.creator35lwb-web/verifimind-genesis)
This MCP server provides X-Z-CS AI validation with auditable reasoning. It reports 8 active tools and 5 tools in maintenance, for a total of 13 tools. Version v0.5.62 is referenced in the repository excerpt, and the description emphasizes structured multi-LLM critique.
🛠️ Key Features
X-Z-CS AI validation with auditable reasoning
Structured multi-LLM critique via specialized agents
Three agents: Innovation, Ethics, Security
Multi-vendor model support (Gemini, Claude, GPT, Groq, Cerebras, Mistral, Ollama)
🚀 Use Cases
Reviewing a concept with separate Innovation, Ethics, and Security perspectives
Generating auditable reasoning as part of AI validation
Critiquing ideas using multiple LLM vendors
⚡ Developer Benefits
MCP-native server and tool-based workflow (13 total tools)
Clear tool lifecycle status: 8 active, 5 in maintenance
Supports multiple LLM providers
⚠️ Limitations
Some functionality is unavailable or undergoing changes (5 tools in maintenance)
Captured live from the server via tools/list.
consult_agent_x
Consult X Intelligent agent for innovation and strategy analysis.
X Intelligent specializes in:
- Innovation potential assessment
- Strategic value analysis
- Market opportunity identification
- Competitive positioning
- Growth potential evaluation
BYOK (v0.4.5): Pass llm_provider and api_key to use your own LLM.
If only api_key is provided, the provider is auto-detected from the key prefix.
Keys are ephemeral (never stored) and garbage collected after the call.
Consult Z Guardian agent for ethical review and Z-Protocol enforcement.
Z Guardian specializes in:
- Ethical implications assessment
- Privacy and data protection review
- Bias and fairness analysis
- Social impact evaluation
- Z-Protocol compliance verification
Z Guardian has VETO POWER. If veto_triggered is True, the concept
should not proceed as it crosses ethical red lines.
BYOK (v0.4.5): Pass llm_provider and api_key to use your own LLM.
If only api_key is provided, the provider is auto-detected from the key prefix.
Keys are ephemeral (never stored) and garbage collected after the call.
Consult CS Security agent for security validation and Socratic interrogation.
CS Security specializes in:
- Security vulnerability assessment
- Attack vector identification
- Data security review
- System integrity analysis
- Socratic questioning (challenging assumptions)
BYOK (v0.4.5): Pass llm_provider and api_key to use your own LLM.
If only api_key is provided, the provider is auto-detected from the key prefix.
Keys are ephemeral (never stored) and garbage collected after the call.
Parameters8
concept_name
string
required
Short name or title of the concept
concept_description
string
required
Detailed description of the concept
context
any
optional
Optional additional context or background
prior_reasoning
any
optional
Optional reasoning from X and Z agents to consider
Optional API key for the provider (ephemeral, never stored)
user_uuid
any
optional
Raw schema
{
"type": "object",
"properties": {
"concept_name": {
"type": "string",
"description": "Short name or title of the concept"
},
"concept_description": {
"type": "string",
"description": "Detailed description of the concept"
},
"context": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional additional context or background"
},
"prior_reasoning": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional reasoning from X and Z agents to consider"
},
"detail": {
"default": "standard",
"type": "string"
},
"llm_provider": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional LLM provider override ('groq', 'anthropic', 'openai', 'gemini', 'mistral', 'ollama', 'mock')"
},
"api_key": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional API key for the provider (ephemeral, never stored)"
},
"user_uuid": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null
}
},
"required": [
"concept_name",
"concept_description"
],
"additionalProperties": false
}
run_full_trinity
Run complete X → Z → CS Trinity validation with Chain of Thought.
This tool orchestrates all three agents in sequence:
1. X Intelligent analyzes innovation and strategy
2. Z Guardian reviews ethics (sees X's reasoning)
3. CS Security validates security (sees X and Z reasoning)
4. Results are synthesized into a unified assessment
Each agent sees the reasoning of previous agents, enabling
true collaborative analysis with full transparency.
BYOK (v0.4.5): Pass llm_provider/api_key for all agents, or use
per-agent overrides (x_provider/x_api_key, z_provider/z_api_key,
cs_provider/cs_api_key). Per-agent params take priority over global.
Keys are ephemeral (never stored) and garbage collected after the call.
Parameters14
concept_name
string
required
Short name or title of the concept
concept_description
string
required
Detailed description of the concept
context
any
optional
Optional additional context or background
save_to_history
boolean
optional
Whether to save the full result to validation history
(default: False). The store is shared and instance-local, retains at
most the 20 newest opt-in results, evicts oldest entries on every
read/write, and clears when the instance is replaced. It has no fixed
time-retention guarantee. Leave False for private or sensitive concepts.
During security maintenance a separately supplied user_uuid has no
effect: no UUID-keyed validation history is written for it.
detail
string
optional
Reasoning verbosity (v0.5.44) — "standard" (default) returns the
auditable `reasoning` block (per-step reasoning, ethics scoring breakdown
+ framework citations, Socratic questions, threat assessment) alongside
the scores; "full" adds per-step evidence and the heaviest structured
fields (12-dimension matrix, 6-stage record, MACP assessment); "summary"
omits the reasoning block for the smallest payload. The block is additive —
existing response fields are unchanged at every level.
llm_provider
any
optional
Optional global LLM provider for all agents
api_key
any
optional
Optional global API key for all agents (ephemeral, never stored)
x_provider
any
optional
Optional provider override for X agent only
x_api_key
any
optional
Optional API key override for X agent only
z_provider
any
optional
Optional provider override for Z agent only
z_api_key
any
optional
Optional API key override for Z agent only
cs_provider
any
optional
Optional provider override for CS agent only
cs_api_key
any
optional
Optional API key override for CS agent only
user_uuid
any
optional
Raw schema
{
"type": "object",
"properties": {
"concept_name": {
"type": "string",
"description": "Short name or title of the concept"
},
"concept_description": {
"type": "string",
"description": "Detailed description of the concept"
},
"context": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional additional context or background"
},
"save_to_history": {
"default": false,
"type": "boolean",
"description": "Whether to save the full result to validation history\n(default: False). The store is shared and instance-local, retains at\nmost the 20 newest opt-in results, evicts oldest entries on every\nread/write, and clears when the instance is replaced. It has no fixed\ntime-retention guarantee. Leave False for private or sensitive concepts.\nDuring security maintenance a separately supplied user_uuid has no\neffect: no UUID-keyed validation history is written for it."
},
"detail": {
"default": "standard",
"type": "string",
"description": "Reasoning verbosity (v0.5.44) — \"standard\" (default) returns the\nauditable `reasoning` block (per-step reasoning, ethics scoring breakdown\n+ framework citations, Socratic questions, threat assessment) alongside\nthe scores; \"full\" adds per-step evidence and the heaviest structured\nfields (12-dimension matrix, 6-stage record, MACP assessment); \"summary\"\nomits the reasoning block for the smallest payload. The block is additive —\nexisting response fields are unchanged at every level."
},
"llm_provider": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional global LLM provider for all agents"
},
"api_key": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional global API key for all agents (ephemeral, never stored)"
},
"x_provider": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional provider override for X agent only"
},
"x_api_key": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional API key override for X agent only"
},
"z_provider": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional provider override for Z agent only"
},
"z_api_key": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional API key override for Z agent only"
},
"cs_provider": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional provider override for CS agent only"
},
"cs_api_key": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional API key override for CS agent only"
},
"user_uuid": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null
}
},
"required": [
"concept_name",
"concept_description"
],
"additionalProperties": false
}
list_prompt_templates
List available prompt templates with optional filtering.
v0.4.0 Unified Prompt Templates feature.
Parameters4
agent_id
any
optional
Filter by agent (X, Z, CS, or 'all')
category
any
optional
Filter by category (startup, security, ethics, etc.)
tags
any
optional
Comma-separated tags to filter by (e.g., 'genesis:phase-1,default')
Import a prompt template from a URL (GitHub Gist, raw file, etc.).
v0.4.0 Unified Prompt Templates feature.
Supports:
- GitHub Gist URLs
- Raw GitHub file URLs
- Any HTTPS URL pointing to JSON/YAML template
Parameters3
url
string
required
URL to import template from
validate
boolean
optional
Whether to validate template content (default: True)
Get statistics about the template registry.
v0.4.0 Unified Prompt Templates feature.
Returns:
Statistics including template counts by agent, phase, and type
Create a structured MACP v2.5 handoff record.
TEMPORARILY DISABLED — this tool is contained pending a security repair
(incident VM-IR-2026-07-28-COORD-01) and always returns a denial.
Handoff bodies written through this tool were stored in a shared namespace
that any unauthenticated caller could read. Nothing is stored or returned
while containment is in force. Coordination state belongs in your own
repository until private, owner-scoped storage ships.
All arguments are accepted for schema stability and are ignored; no
supplied value is stored, logged, or echoed back.
Returns:
A COORDINATION_TEMPORARILY_DISABLED error with an incident reference.
Read the most recent coordination handoff record(s).
TEMPORARILY DISABLED — this tool is contained pending a security repair
(incident VM-IR-2026-07-28-COORD-01) and always returns a denial.
This read path returned complete handoff bodies from a caller-selected
namespace with no authenticated ownership check. It is closed to every
caller — with or without a key — while owner-scoped access control is
rebuilt. No record body is returned.
All arguments are accepted for schema stability and are ignored.
Returns:
A COORDINATION_TEMPORARILY_DISABLED error with an incident reference.
Return current team coordination state and session summary.
TEMPORARILY DISABLED — this tool is contained pending a security repair
(incident VM-IR-2026-07-28-COORD-01) and always returns a denial.
This summary exposed agent identities, pending actions, open blockers, and
a timestamped activity index from a caller-selected namespace with no
authenticated ownership check — enough to enumerate a namespace before
reading it. It is closed to every caller while owner-scoped access control
is rebuilt.
All arguments are accepted for schema stability and are ignored.
Returns:
A COORDINATION_TEMPORARILY_DISABLED error with an incident reference.
Accounts: registration and UUID-linked account features (including the personal usage dashboard) are temporarily unavailable during security maintenance. You do not need an account today — every active tool works anonymously at the standard rate limit. Advance notice: under the published Terms and Privacy Policy, from October 20, 2026 the four execution tools (consult_agent_x, consult_agent_z, consult_agent_cs, run_full_trinity) will require a free registered account. Tool discovery and the template-read tools stay open without registration, and the tools remain free.
What this is
VerifiMind PEAS is an MCP server that runs your concept through three specialized LLM judges in sequence:
Agent
Role
Question it answers
X (Innovation)
Innovation & competitive positioning
"Is this novel? What's the prior art? What's the strategic angle?"
"What risks does this raise? GDPR, EU AI Act, SG MGF, etc."
CS (Security)
Security validation, OWASP Agentic AI Top 10
"What can break? What's the attack surface? What's the reasoning-layer audit say?"
Each later stage receives the prior completed stage's structured analysis. You get a unified assessment with scores, findings, recommendations (PROCEED / REVISE / REJECT), and explicit stage provenance. Only documented output fields are passed forward; hidden model internals are not exposed.
What this is not: "Verification" in the formal-methods sense. The output is structured multi-LLM critique, not a mathematical proof. We make that distinction explicitly.
The 13 tools
All 13 tools remain free for everyone under the Core Tools Always Free pledge. 8 are currently active; 5 are temporarily unavailable for security maintenance — three coordination tools plus two custom-template mutation tools.
Trinity validation (4 tools)
consult_agent_x — Innovation analysis with competitive positioning
consult_agent_z — Ethics review with 21-framework jurisdictional coverage
consult_agent_cs — Security validation, OWASP Agentic AI Top 10
run_full_trinity — X → Z → CS pipeline with structured stage context and a unified assessment
Output-integrity contract (v0.5.58): an agent's generated scores, findings,
recommendation, analysis, and veto state are returned only when its inference
quality is explicitly real. Provider failures now degrade only the affected
stage, preserve successful sibling stages, and expose typed stage attribution.
Incomplete runs withhold aggregate score/confidence and are not saved to shared
history.
import_template_from_url — temporarily unavailable pending owner isolation and URL-fetch hardening
get_template_statistics — Registry stats by agent / phase / type
Coordination (3 tools) — TEMPORARILY UNAVAILABLE
coordination_handoff_create — disabled
coordination_handoff_read — disabled
coordination_team_status — disabled
These three tools are disabled and currently return COORDINATION_TEMPORARILY_DISABLED for every caller.
Records created through them were stored in a shared, unauthenticated namespace; they are no longer readable or writable through the public API. The four validation tools remain fully available.
They will return only after private, owner-scoped storage ships. Until then, keep coordination state in your own repository — the handoff markdown format is documented in this repo. Incident reference: VM-IR-2026-07-28-COORD-01.
The Always Free pledge is unchanged. This is a security containment, not a paywall and not a tier change. Nothing here is or becomes a paid feature.
register_custom_template and import_template_from_url return CUSTOM_TEMPLATE_TEMPORARILY_DISABLED for every caller. Public built-in template listing, retrieval, export, and statistics remain available and exclude process-local custom entries.
This containment prevents cross-caller custom-template visibility while owner-scoped storage is built and removes the public arbitrary URL-fetching path. Incident reference: VM-IR-2026-08-01-TEMPLATE-01.
Core Tools Always Free Pledge
All VerifiMind PEAS validation tools are free to use, forever. No paywall, no premium tier for tool access. Rate limits apply for system health only (not as monetization). Paid services, when they launch, will be consultation reports — separate from the tools.
Ratified by L (CEO) + Alton (Human Orchestrator) + T (CTO) on May 9, 2026. Active in production since v0.5.28 (May 10, 2026).
Rate limits (system health, equal for all tiers):
Tier
Identity
Limit
Anonymous
IP only
10 req/60s
Scholar
UUID (free registration)
30 req/60s
EA / PILOT
UUID + email
100 req/60s
While account features are in maintenance, every caller is served at the anonymous limit.
Methodology overview
VerifiMind PEAS productizes the multi-judge LLM evaluation pattern — a well-established approach in the AI evaluation literature — into an opinionated MCP server with three specialized roles, a Genesis Master Prompt continuity layer, and a multi-vendor BYOK architecture.
What's ours:
Productization quality of the X / Z / CS specialization
MCP-native exposure (works in Claude Code / Cursor / VS Code / ChatGPT Codex)
Multi-vendor design (not locked to one LLM family)
Genesis Master Prompt — stateful continuity across multi-model workflows
21-framework jurisdictional coverage in the Ethics agent (GDPR · EU AI Act · SG MGF · etc.)
What's prior art: Multi-judge LLM evaluation, LLM-as-judge scoring, multi-model orchestration. See Related Work for citations.
We do not claim the underlying methodology is novel.
Architecture
The full system architecture — from the foundational X / Z / CS multi-agent validation design through the Phase 90 production deployment (MCP transport, BYOK provider layer, security hardening, FLYWHEEL coordination) — is documented in a single comprehensive, fact-checked diagram:
Release: merge 2685d4c7 · Cloud Build 50896a37-f06e-484a-beb8-6647435f0eb6 · revision verifimind-mcp-server-00513-klm · post-deploy Trinity: one anonymous run, X/Z/CS = real/real/real, complete (CS used 3,204 of its 3,740-token Groq completion reservation, so the Z/CS headroom limitation recorded for v0.5.63 remains); see SERVER_STATUS.md
Authentication (v0.5.63): a native OAuth 2.1 authorization server is deployed dark — credential issuance and MCP enforcement are both off, so every tool that was anonymous stays anonymous
Tools: 13 defined / 8 active / 5 temporarily unavailable (all remain free)
Providers: 6 remote BYOK catalogues (Gemini · Anthropic · OpenAI · Groq · Cerebras · Mistral) plus caller-managed local Ollama
Hosted routing: X = Gemini gemini-3.5-flash-lite; Z/CS = Groq openai/gpt-oss-120b
MCP Registry package:3.41.0 — published by the v0.5.64 GitHub Release on September 24, 2026, API-verified live (tag at the exact release merge 2685d4c7)
For live release history, see /changelog. We deliberately do not display unaudited "total users" numbers because bot, scanner, and development traffic can distort them. Exact deployment provenance and the full dated operational snapshot are documented in SERVER_STATUS.md.
Common mistakes
Mistake
Fix
Using https://verifimind.ysenseai.org/mcp (no slash)
Use /mcp/ with trailing slash — required by streamable-http transport
Connecting via server.smithery.ai/...
Smithery legacy was sunset March 1, 2026. Use the direct URL above.
Expected security containment. Use built-in templates or keep custom templates in your own repository until owner-scoped storage ships.
Trying to call coordination tools and seeing "PIONEER_TIER_REQUIRED"
You're on v0.5.27 or older — the paywall was removed in v0.5.28 (May 10, 2026). Tool access is free; note the 3 coordination tools are separately disabled (row above).
If you use VerifiMind PEAS in research or a project, please cite. We'd love to hear about it — open a GitHub Discussion.
VerifiMind PEAS (server)
bibtex
@software{verifimind_peas_2026,
author = {Lee, Alton and {Manus AI} and {Claude Code}},
title = {VerifiMind PEAS: Multi-Agent AI Validation MCP Server},
year = {2026},
url = {https://github.com/creator35lwb-web/VerifiMind-PEAS},
doi = {10.5281/zenodo.17980791},
note = {Multi-vendor MCP server for structured multi-LLM critique}
}
VerifiMind PEAS is released under the MIT License. See LICENSE for the full text.
code
Copyright (c) 2025-2026 Alton Lee Wei Bin (creator35lwb)
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
The methodology is freely usable under MIT. Forks and derivatives must use different branding.
For paid consultation engagements (planned, not yet active), use GitHub Discussions or email — we'll publish service details and pricing when they're ready.
Acknowledgments
VerifiMind PEAS was built collaboratively by the FLYWHEEL TEAM — a human orchestrator working with multiple AI agents (Manus AI, Claude Code, Perplexity, Antigravity/Gemini, GodelAI). Multi-agent coordination uses the open MACP protocol (v2.5 "Loop Engineering", published 2026-07-14).
External Model Council review (Claude Opus 4.7 + GPT-5.5 + Gemini 3.1 Pro, May 9, 2026) shaped the current positioning. See docs/case-studies for application examples.
Last Updated: September 24, 2026 · Version: v0.5.64 · MCP Registry: 3.41.0 · MACP: v2.5 "Loop Engineering" (DOI)
