veroq-mcp

Verified AI. One function call. Stop shipping hallucinations.

62 MCP tools. shield() any LLM output. Evidence chains on every claim. Finance as flagship vertical, with legal, research, compliance, and custom domains built in.

veroq.ai | API Reference | Pricing

Why VeroQ?

Tools (62)

Hero Tools

High-Level Tools

Swarm & Runtime

Market Data (16 tools)

veroq_ticker_price (free), veroq_full, veroq_candles, veroq_technicals, veroq_earnings, veroq_market_movers, veroq_market_summary, veroq_economy, veroq_forex, veroq_commodities, veroq_sectors, veroq_portfolio_feed, veroq_screener, veroq_screener_presets, veroq_alerts, veroq_ticker_score

Crypto (4 tools)

veroq_crypto, veroq_crypto_chart, veroq_defi, veroq_defi_protocol

Search & Intelligence (15 tools)

veroq_search, veroq_feed, veroq_brief, veroq_extract, veroq_entities, veroq_trending, veroq_compare, veroq_research, veroq_timeline, veroq_forecast, veroq_contradictions, veroq_events, veroq_diff, veroq_web_search, veroq_crawl

Fundamentals (5 tools)

veroq_insider, veroq_filings, veroq_analysts, veroq_congress, veroq_institutions

Other (7 tools)

veroq_ticker_news, veroq_ticker_analysis, veroq_search_suggest, veroq_social_sentiment, veroq_social_trending, veroq_ipo_calendar, veroq_run_agent

Reports (2 tools)

veroq_generate_report, veroq_get_report

Enterprise

Enterprise customers get:

• Permission engine — allow/deny/review rules with wildcard patterns
• Decision lineage — full rule evaluation trace for every tool call
• Escalation — high-stakes queries trigger human-in-the-loop review
• Audit trails — every decision logged with session tracking
• External MCP — securely proxy external APIs with trust levels and rate limits
• Feedback loop — self-improvement with web search fallback
• Observability — admin dashboard with verification stats, safety triggers, credit consumption

Contact enterprise@veroq.ai or visit veroq.ai/pricing.

Environment Variables

POLARIS_API_KEY and POLARIS_BASE_URL accepted for backward compatibility.

TradingAgents-Pro

TradingAgents-Pro is a demo showcasing what VeroQ can do — 15 agents running on the verified intelligence layer. VeroQ is the product. TradingAgents-Pro is the showcase.

License

MIT

Install

npm install -g veroq-mcp

Get your API key at veroq.ai/settings.

Quick Start

Once installed, two tools handle most use cases:

veroq_ask      "How is NVIDIA doing?"          → price, technicals, earnings, sentiment, news
veroq_verify   "Apple beat Q4 earnings by 20%"  → SUPPORTED (92%), 5 sources, evidence chain

For multi-agent workflows, use the Verified Swarm:

veroq_run_verified_swarm   { "query": "Analyze NVDA for a long position" }
→ 5 agents: planner → researcher → verifier → critic → synthesizer
→ Every step verified, escalation on high-stakes, full decision lineage

For domain-specific pipelines, use the Agent Runtime:

veroq_create_runtime   { "vertical": "finance", "query": "Analyze NVDA" }
veroq_create_runtime   { "vertical": "legal", "query": "GDPR data retention" }

IDE Setup

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "veroq": {
      "command": "veroq-mcp",
      "env": { "VEROQ_API_KEY": "your-api-key" }
    }
  }
}

Cursor

Settings > MCP Servers > Add Server: Name veroq, Command veroq-mcp, Env VEROQ_API_KEY=your-api-key.

Programmatic Usage

VeroQ is also available as a TypeScript/Python library:

import { createRuntime, registerExternalMcpServer } from "veroq-mcp";

// Finance runtime with external market data
const runtime = createRuntime({
  vertical: "finance",
  enterpriseId: "acme-capital",
  costMode: "balanced",
  enableSelfImprovement: true,
  externalServers: [{
    serverId: "bloomberg",
    serverUrl: "https://api.bloomberg.com",
    auth: { type: "bearer", credential: process.env.BLOOMBERG_TOKEN },
    allowedTools: ["get_security", "get_analytics"],
    trustLevel: "read-only",
  }],
});

const result = await runtime.run("Analyze NVDA for a long position");
console.log(result.synthesis.summary);
console.log(result.budget);           // { spent: 12, remaining: 38 }
console.log(result.verificationSummary); // { avgConfidence: 82, flaggedSteps: 0 }

Verify Any LLM Output

// Extract and verify every claim in arbitrary LLM text
const verified = await client.verifyOutput("NVIDIA's Q4 revenue exceeded $22B and margins expanded to 75%");
console.log(verified.claims);           // each claim with verdict, confidence, correction
console.log(verified.overall_confidence);

Agent Memory

// Store context tied to an agent
await client.memoryStore({ agent_id: "my-bot", key: "nvda-thesis", value: "bullish on AI capex" });

// Recall relevant context for a query
const context = await client.memoryRecall({ agent_id: "my-bot", query: "NVDA outlook" });

// List all stored memories
const memories = await client.memoryList({ agent_id: "my-bot" });

Real-Time Verification Stream

// Persistent SSE — pushes ticker_status, signal_change, confidence_change, claim_update
const stream = client.watch({ tickers: ["NVDA", "AAPL"], agent_id: "my-bot" });
stream.on("signal_change", (event) => console.log(event));