--- title: Collective Intelligence Orchestrator emoji: 🐢 colorFrom: purple colorTo: indigo sdk: gradio sdk_version: 5.33.1 app_file: app.py pinned: true license: apache-2.0 tags: - mcp - agents - gradio - hackathon - agent-demo-track - collective-intelligence - mcp-server-track short_description: Gradio-based interface coordinates a network of agents --- # 🧠 Collective Intelligence Orchestrator This Gradio-based interface coordinates a network of autonomous AI agents using Hugging Face's MCP protocol. Collective Intelligence Orchestrator is a Gradio-powered interactive application that coordinates a swarm of autonomous AI agents using the Hugging Face Model Context Protocol (MCP). It simulates real-time, cross-domain collaboration between specialized agents for responding to real-world challenges. This project is built for the Hugging Face + Open Source AI Hackathon and explores how autonomous model-to-model communication can drive situational understanding, policy modeling, and collective action at scale. # 🚀 What It Does You provide a real-world scenario — such as a humanitarian crisis, environmental disruption, or geopolitical event. The orchestrator routes this context to a network of specialized MCP-compliant agents. These agents independently process the scenario and return responses from their unique domain lenses. The orchestrator aggregates their outputs and presents a coordinated, multi-perspective analysis. ## 🤖 Agents in the Swarm: Each agent specializes in a unique area of knowledge. Current agents include: - Climate Sensor - Policy Modeler - Economic Forecast - Media Monitor - Public Health - NGO Matcher Submit a real-world scenario, and watch the swarm collaborate in real time. All agents comply with MCP and are auto-discoverable, allowing seamless plug-and-play integration of new models. ## 📊 Test Case Example: Input: - “Massive flooding in coastal Bangladesh, affecting over 200,000 residents. Power outages, displacement, and rising waterborne diseases reported.” Expected output: - Climate Sensor: Critical anomaly detected. - Policy Modeler: Emergency zoning, sanitation relief, medical deployment. - Economic Forecast: Estimated GDP loss, recovery timeline. - Media Monitor: High sentiment panic, moderate misinfo risk. - Public Health: Cholera risk, hospital overload. - NGO Matcher: Suggest 2–3 orgs that can help. ## 🛠 Architecture - Frontend: Gradio interface for input and real-time orchestration feedback - Orchestrator Logic: app.py routes user scenarios to agents based on roles and context - Agent Discovery: MCP auto-discovers agent endpoints via metadata (.well-known/mcp.yaml) - Model Interfacing: Each agent may wrap a hosted LLM (Claude, GPT-4, etc.) or a local model with domain-aligned prompts Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference