Give your AI agent a 120-tool global intelligence feed

What it does

The project exposes 120 async tools across 30-odd intelligence domains—financial markets, SEC filings, armed conflict, military aircraft tracking, cyber threats, climate anomalies, and maritime activity—to any MCP client. All data is pulled from free public APIs, so there are no subscription fees or API keys to manage. A background collector aggregates everything into a Qdrant vector store, letting agents run semantic searches like “military activity near Taiwan” across accumulated history.

The interesting bit

Data resilience is treated as a first-class concern: every source gets its own circuit breaker, stale-data fallback, and per-feed rate limiting. It also generates cited situation briefs and daily digests, synthesizing cross-domain signals through local Ollama with a mechanically-cited fallback when the LLM is unavailable.

Key highlights

• 120 MCP tools spanning finance, geopolitics, defense, cyber, climate, space weather, and infrastructure
• Entirely free data layer: Yahoo Finance, USGS, ACLED, SEC EDGAR, NOAA, NASA FIRMS, and dozens more public APIs
• Qdrant vector store with FastEmbed for natural-language search across historical intelligence
• Live ops-center dashboard with Leaflet map layers, 47 SSE feeds, and per-source health circuit breakers
• User-defined geofences with multi-domain hotspot escalation scoring

Caveats

• PDF report generation requires native system dependencies (pango on macOS)
• Full AI synthesis depends on a local Ollama instance; without it, briefs fall back to mechanical citation lists
• Python 3.11+ required

Verdict

Worth a look if you’re building agents that need ground-truth world awareness without racking up SaaS bills. Skip it if you want polished proprietary intelligence or a fully managed cloud service.