PROVNAI Ecosystem#
The PROVNAI initiative provides the trust layer for the agentic future. Below is the unified Nexus for all core repositories, protocols, and forensic tools in the ecosystem.
Integration-Ready Templates#
Pre-configured deployment and evaluation paths for testing core PROVNAI ecosystem components with minimal overhead.
Deploy VEX on Railway
Bring up a VEX evaluation instance with the right port mapping, health checks, and deployment caveats.
McpVanguard on Railway
Deploy the current McpVanguard gateway on Railway and attach a remote MCP client with the right auth headers.
Agent Execution Queue
Create an agent, queue work, and poll the resulting job with exact request and response examples.
Verifiable Pipeline Audit Trail
Generate valid and tampered audit exports, then verify them with the public CLI.
📦 Repositories#
🛡️ vex#
VEX Protocol — The trust layer for AI agents. Adversarial verification, temporal memory, Merkle audit trails, and tamper-proof execution. Built in Rust.
vex-core
Core types including Fractal Agents, Genome, Merkle Trees, and Audit Logs.
vex-router
Intelligent LLM routing with cost/latency/quality optimization and adversarial detection.
vex-adversarial
Red/Blue agent debate engine and weighted consensus protocols for verification.
vex-temporal
Bio-inspired episodic memory with automatic decay and vectorized permanent storage.
vex-persist
SQLite-backed persistence for agents, contexts, audit logs, and embeddings.
vex-anchor
Multi-chain anchoring layer supporting Ethereum, Celestia, and OpenTimestamps.
vex-algoswitch
Dynamic runtime algorithm selection and optimization engine for VEX.
vex-runtime
Agent orchestrator and executor with built-in adversarial verification loops.
vex-llm
Provider-agnostic LLM interface with support for WebSocket, MCP, and circuit breakers.
vex-api
Flagship HTTP API server with JWT authentication and A2A protocol support.
vex-queue
Distributed background job processing for high-throughput agent operations.
vex-cli
Command-line interface for managing agents, verifying logs, and anchoring proofs.
vex-macros
Procedural macros for ergonomic agent definition and DNA mapping.
vex-chora
Decentralized governance and protocol state management.
vex-hardware
Hardware-level security and trusted execution environment (TEE) integration.
vex-server
High-performance P2P networking and gossip protocol for VEX agents.
vex-sidecar
L7 proxy and observability sidecar for monitoring agent communication.
🛡️ McpVanguard#
McpVanguard — Active firewall for the Model Context Protocol (MCP). It acts as a real-time 'Reflex System' between AI agents and their tools. Built in Python.
🔍 vex-explorer#
VEX Explorer — Forensic tool for verifying the mathematical integrity of VEX Evidence Packets. Powered by a high-performance Rust core (WASM).
🐍 vex-sdk#
VEX SDK (Python) — Official Python client for building autonomous VEX fractal agents.
🧪 provnai#
ProvnAI — Open Research Initiative building the immune system for AI. Verifiable trust layers for autonomous agents.
🔑 provn-sdk#
provn-sdk — Client-side SDK for privacy-preserving cryptographic signing. Hash and sign claims locally — raw data never leaves your device. Built in Go/Rust.
📉 vex-halt#
VEX-HALT — Benchmark suite for AI verification systems. 443+ tests for calibration, robustness, honesty, and proof integrity.
🛡️ provnzero#
ProvnZero — Zero Knowledge. Zero Retention. Zero Compromise. High-performance ZDR proxy for AI APIs.
provnzero-proxy
Rust-based Zero Data Retention proxy with OS-level secure buffers.
provnzero-sdk
TypeScript SDK for client-side HPKE payload sealing.