Decision Support Network

Inspiration

What it does

How I built it

Challenges I ran into

Accomplishments that I'm proud of

What I learned## Inspiration

As AI agents become the primary way people interact with web services, we realized that individual decision-support tools are hard to discover. An agent shouldn't need to know about 12 specialized tools upfront - it should be able to ask one discovery layer and get routed to the right one.

What it does

The Decision Support Network (DSN) is a WebMCP-enabled discovery and routing hub for 12 specialized AI decision-support tools spanning healthcare, legal, housing, energy, education, and more.

DSN demonstrates a two-hop agent workflow:

1. An AI agent discovers DSN and uses its WebMCP tools to find the right decision-support capability
2. DSN routes the agent to a second, specialized WebMCP-enabled Decision Support App (DSA)
3. The agent calls the specialized tool directly for structured analysis

WebMCP Tools:

• find_decision_tools - Natural-language intent matching
• search_tools - Keyword search across all registered tools
• list_categories - Browse tools by category

Four Access Interfaces (one canonical routing engine):

• Browser WebMCP via document.modelContext.registerTool()
• Remote MCP (JSON-RPC 2.0) at /api/mcp
• REST API with full OpenAPI 3.1 spec
• Human UI at https://decisionsupportnetwork.com

How we built it

Built with Next.js, TypeScript, Prisma/PostgreSQL, and NextAuth v5. The routing engine uses LLM-backed intent matching with keyword fallback for deterministic parity across all four access surfaces.

What's next

Expanding the network with more decision-support tools and enabling agents to chain multiple tools for complex multi-step decision workflows.

What's next for Decision Support Network

Built With

• https://decisionsupportnetwork.com
• next.jstypescriptwebmcpmcpprismapostgresqltailwind