{"slug": "che-mcp-building-argentina-s-first-national-mcp-ecosystem-5-stage-classifier-wma", "title": "CHE MCP — Building Argentina's First National MCP Ecosystem: 5-Stage Classifier, WMA Online Learning, 748 Datasets", "summary": "A developer in Bahía Blanca, Argentina built CHE MCP, the country's first national MCP ecosystem that connects AI agents to 80+ official Argentine data sources through a single server. The system uses a 5-stage classifier with a Weighted Majority Algorithm for online learning, achieving 95.45% accuracy on MCPAgentBench, and serves 748 Parquet datasets with natural language to SQL conversion.", "body_md": "Argentina just got its first national MCP ecosystem — and it was built from Bahía Blanca.\n\nCHE MCP is an intelligent gateway that connects any AI agent with real-time Argentine data. Dollar exchange rates, weather, football, tax compliance (ARCA), inflation, public transit — 80+ official data sources through a SINGLE MCP server.\n\nWhy does this matter? Because right now, if you want your AI to answer \"¿cuánto está el dólar blue?\", you either Google it yourself or install 80 different MCP servers. CHE MCP solves that with a gateway that understands natural language in Spanish and routes queries automatically.\n\nQuery: \"dolar blue hoy\"\n\n│\n\n┌────▼─────┐ Stage 1 — Keyword matching\n\n│ Keyword │ 3,000+ keywords across 182 classified domains\n\n└────┬─────┘\n\n│\n\n┌────▼─────┐ Stage 2 — WMA weighted routing\n\n│ WMA │ Weighted Majority Algorithm: learns from every query\n\n└────┬─────┘\n\n│\n\n┌────▼─────┐ Stage 3 — Semantic embeddings\n\n│ Embedding │ 384-dim vectors (all-MiniLM-L6-v2) with Jaccard fallback\n\n└────┬─────┘\n\n│\n\n┌────▼─────┐ Stage 4 — Data Node search\n\n│ Data Node │ DuckDB SQL over 748 Parquet datasets + NL-to-SQL\n\n└────┬─────┘\n\n│\n\n┌────▼─────┐ Stage 5 — LLM fallback\n\n│ LLM │ External endpoint (optional, configurable)\n\n└────┬─────┘\n\n│\n\n┌────▼─────┐\n\n│ Response │ \"Dólar blue: $1,245 / $1,265 compra/venta\"\n\n└──────────┘\n\nThe Weighted Majority Algorithm (WMA) is an online learning system embedded directly in the router. Every domain starts with equal weight (1.0). When a query succeeds, the winning domain gets reinforced (+0.1). When it fails, the domain gets penalized (−0.1). Weights are bounded at [0.1, 5.0] and persisted to disk — the router starts warm and improves with every query.\n\n**Benchmark: 95.45% Top-First-Score accuracy** on MCPAgentBench (66 diverse queries).\n\n748 Parquet datasets from datos.gob.ar (Argentina's open data portal), compressed 9.92× with Zstd (404 MB vs 3.92 GB CSV). The Data Node converts natural language to SQL:\n\nUser: \"¿Cuánto aumentó la inflación en 2024?\"\n\n→ DuckDB generates: SELECT AVG(valor) FROM indice_precios_consumidor\n\nWHERE fecha BETWEEN '2024-01-01' AND '2024-12-31'\n\n→ Result: 117.8% anual\n\nSQL injection guardrails, read-only enforcement, 5-second timeout, 1,000-row result limit.\n\n| Pattern | Implementation |\n|---|---|\n3-tier cache |\nIn-memory LRU (200 entries) → disk (atomic writes) → live CKAN |\nCircuit breaker |\nPer-dataset, 3-failure threshold, 60s cooldown, serves stale data |\nRequest collapsing |\nConcurrent identical queries share a single upstream fetch |\nPredictive pre-fetch |\nTop-10 hot datasets refresh every 15 minutes |\nRate limiting |\nToken bucket per API key, 100 req/min, noisy neighbor isolation |\n\nThe Model Context Protocol is undergoing its biggest architectural update in July 2026 — mandatory Streamable HTTP transport, stateless architecture. CHE MCP was architected for this from day one:\n\nBuilt from Bahía Blanca, Argentina 🇦🇷 with [Gentle AI](https://github.com/Gentleman-Programming/gentle-ai)'s SDD orchestration + [Engram](https://github.com/Gentleman-Programming/engram) persistent memory.\n\nFull technical documentation: [github.com/Albano-schz/che-mcp-docs](https://github.com/Albano-schz/che-mcp-docs)\n\n*What questions do you have about building MCP ecosystems at national scale?*", "url": "https://wpnews.pro/news/che-mcp-building-argentina-s-first-national-mcp-ecosystem-5-stage-classifier-wma", "canonical_source": "https://dev.to/albanoschz/asdas-3107", "published_at": "2026-06-22 02:32:58+00:00", "updated_at": "2026-06-22 02:39:28.962259+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "natural-language-processing", "ai-agents", "developer-tools"], "entities": ["CHE MCP", "Gentle AI", "Engram", "DuckDB", "MCPAgentBench", "datos.gob.ar", "Albano-schz", "Bahía Blanca"], "alternates": {"html": "https://wpnews.pro/news/che-mcp-building-argentina-s-first-national-mcp-ecosystem-5-stage-classifier-wma", "markdown": "https://wpnews.pro/news/che-mcp-building-argentina-s-first-national-mcp-ecosystem-5-stage-classifier-wma.md", "text": "https://wpnews.pro/news/che-mcp-building-argentina-s-first-national-mcp-ecosystem-5-stage-classifier-wma.txt", "jsonld": "https://wpnews.pro/news/che-mcp-building-argentina-s-first-national-mcp-ecosystem-5-stage-classifier-wma.jsonld"}}