omni-ai-routing-engine
by Hivanconh
README.md
# Omni-Perspective AI Health & Routing Engine (MCP Server)
[](https://smithery.ai)
[](https://modelcontextprotocol.io)
[](https://workers.cloudflare.com)
[](https://getalby.com)
A high-performance, edge-deployed Model Context Protocol (MCP) server running on Cloudflare Workers. It acts as a predictive decision engine and circuit-breaker orchestrator for AI agents, multi-model consensus systems, and autonomous workflows (**Council System v1**).
---
## đŠī¸ Architecture & Capabilities
- **Low-Latency Telemetry:** Instant edge latency checks via Cloudflare internal routing.
- **Failover & Circuit Breaker:** Dynamically evaluates network degradation to issue rerouting directives (`PROCEED` vs `FAILOVER_TO_SECONDARY`).
- **M2M Lightning Paywall (L402 / x402 Ready):** Integrated with Lightning Network addresses (`admiredhut112@walletofsatoshi.com`) for micro-monetization (1 to 5 SATS per call).
- **`llms.txt` Native:** Exposes machine-readable documentation at `/llms.txt` for autonomous agent discovery.
---
## đ ī¸ Tools Provided
### `get_endpoint_telemetry` (Cost: 1 SAT)
Atomic telemetry check returning live latency and health metrics directly from the edge node.
**Input Parameters:**
- `target_provider` *(string, optional)*: Network provider or node identifier to evaluate (e.g., `cloudflare-edge`).
### `get_health_and_routing` (Cost: 5 SATS)
Predictive decision engine returning failure risk, network degradation, circuit-breaker directives, and M2M failover routing consensus.
**Input Parameters:**
- `target_provider` *(string, optional)*: Target model ID to test latency against (e.g., `openai/gpt-4o`, `anthropic/claude-3-5-sonnet`).
- `risk_threshold` *(string, optional)*: Sensitivity threshold (`STRICT`, `BALANCED`, `PERMISSIVE`).
---
## đ Quick Setup & Integration
### 1. Claude Desktop Integration
Add the following snippet to your `claude_desktop_config.json`:
```json
{
"mcpServers": {
"omni-perspective-engine": {
"command": "npx",
"args": [
"-y",
"@smithery/cli@latest",
"run",
"shrill-wind-8acb.hivanapps.workers.dev",
"--config",
"{}"
]
}
}
}
```
### 2. Cursor / Windsurf Integration
Go to **Settings** -> **MCP Servers** -> **Add New MCP Server**:
- **Type:** `SSE` or `HTTP`
- **URL:** `https://shrill-wind-8acb.hivanapps.workers.dev/mcp`
---
## đ Standard Endpoint Discovery
- **MCP Endpoint:** `https://shrill-wind-8acb.hivanapps.workers.dev/mcp`
- **LLM Context:** `https://shrill-wind-8acb.hivanapps.workers.dev/llms.txt`
---
## đ License
MIT License
This server cannot be deployed
Maintenance
ActivitySlowing
ResponsivenessNo issues