Open Beta — v0.3.1

Runtime trust scoring for MCP tools — before your AI agent calls them.

15,536 MCP servers scored. Real confidence metrics. One API call.

Try the API →
v0.3.1 · Apache-2.0 · API status: checking…
Live API Demo

See it work. Right now.

This isn't a mockup. Hit the button and watch the Resonance Engine score MCP tools in real time.

Request
Response
# Resolve the best MCP tool for an intent
curl -X POST https://api.xluxx.net/resolve-tool \
  -H "Content-Type: application/json" \
  -d '{"intent":"search the web"}'
{
  "best_server": "brave-search",
  "confidence": 0.94,
  "fallback": "tavily-search",
  "risk_flags": [],
  "resonance_score": 0.91,
  "fractal_reliability": 0.96,
  "coherence_drift": 0.02
}

Live response from api.xluxx.net:


  
Use Cases

Who is this for?

If your AI agents call external tools, you need trust scoring.

AI Agent Developers

Building with LangChain, CrewAI, or AutoGen? Add a trust gate before every tool call. Know which MCP server to route to — and which to avoid — before your agent acts.

MCP Tool Builders

Ship trust signals with your server. Show users your reliability score, uptime history, and coherence metrics. Stand out in a crowded registry.

DevOps & AI in Production

Running AI workflows in production? Monitor tool reliability drift, get fallback routing, and catch behavioral changes before they break your pipelines.

Methodology

How the Resonance Engine works

Four independent signals, combined into a single trust score. Every metric is transparent and auditable.

Fractal Reliability Index

Temporal Stability

Multi-scale pattern detection across 1h, 24h, 7d, and 30d windows. Servers that are reliable at every timescale score higher than those with intermittent spikes.

Coherence Drift

Behavioral Baseline

Measures how much a server's behavior has changed from its established baseline. Low drift = predictable. High drift = something changed — investigate.

Toolchain Resonance

Amplification Score

How well does this tool work in combination with others? Tools that amplify their neighbors' reliability score higher than isolated performers.

Fallback Stability

Backup Quality

Is there a reliable alternative? A high fallback stability score means your agent has a safe Plan B if the primary tool degrades.

trust_score = w1·fractal_reliability + w2·(1 - coherence_drift) + w3·toolchain_resonance + w4·fallback_stability

Weights adapt per-query based on intent criticality. Default: w=[0.35, 0.25, 0.20, 0.20]

Live Infrastructure

Platform stats

Pulled live from the XLUXX API on page load.

MCP Servers Tracked
Tools Cataloged
API Status
Version
Pricing

Start free. Scale when ready.

No credit card required for the free tier. Upgrade instantly via Stripe.

Free
$0/mo
  • 10 calls / minute
  • 100 calls / day
  • All scoring metrics
  • Community support
Get API Key
Pro
$149/mo
  • 10,000 calls / day
  • Dedicated support
  • Custom scoring weights
  • SLA guarantee
Subscribe →
Integration

Three lines to trust-scored tools.

SDKs for Python and Node.js. Or just use the REST API.

Python
Node.js
MCP Server
from xluxx import TrustClient
client = TrustClient(api_key="your-key")
result = client.resolve("search the web")
# result.best_server → "brave-search"
print(f"Use {result.best_server} (confidence: {result.confidence})")
import { TrustClient } from 'xluxx-trust';
const client = new TrustClient({ apiKey: 'your-key' });
const result = await client.resolve('search the web');
console.log(result.best_server); // "brave-search"
# claude_desktop_config.json
{ "mcpServers": { "xluxx-trust": { "command": "npx", "args": ["xluxx-trust-mcp"] } } }
Security Intelligence

Passive OSINT audits for every MCP server.

Every server in our registry gets a passive security assessment — no exploitation, no noise. CVE matching, SSH version detection, secret scanning, DNS recon, and OSINT profiling.

Servers Audited
Critical Findings
CVEs Detected
Secrets Exposed
SSH Vulnerabilities
Grade F Servers
Ecosystem Grade
Grade Distribution
A
B
C
D
F
🔍

CVE & OSV Matching

Detected framework versions are cross-referenced against OSV.dev and a curated CVE database. Every match gets a CVSS v3 score and CWE classification.

🔑

Secrets & Code Scanning

Public GitHub repositories are scanned for exposed API keys, tokens, and credentials — Stripe, AWS, OpenAI, Anthropic, GitHub PATs, private keys, and more.

🌐

Network & SSH Recon

Passive TCP port probing detects unexpected exposed services. SSH banners are fingerprinted — OpenSSH versions checked against regreSSHion (CVE-2024-6387) and Terrapin.

🧩

OSINT Profiling

RDAP/WHOIS for registrar and expiry data. Developer username checked across GitHub, npm, Docker Hub, PyPI, GitLab, HuggingFace. Email checked via Gravatar and HaveIBeenPwned.

📧

DNS Hardening

SPF, DMARC, MX and NS records audited. Missing email authentication policies flagged — domain spoofing and phishing risks quantified with CWE-346.

⚖️

CFAA-Compliant

All scans are passive and non-destructive. We read public data — no exploitation attempts, no unauthorized access, no active attacks. Research-grade intelligence within legal bounds.

Leaderboard

Top 25 Most Reliable MCP Servers

Ranked by composite trust score. Updated every hour from live monitoring data.

#ServerTrust ScoreReliabilityDriftCategory
Loading leaderboard from API…