Build on Nynch
42 MCP tools for AI agents to search contacts, manage deals, analyze relationships, and orchestrate multi-agent workflows.
Quick Start
Connect any MCP-compatible AI client to Nynch in three steps.
Get an API Key
In Nynch, go to Settings → MCP Server and create an API key. Choose a scope: read, read-write, or admin.
Configure Your Client
Add the Nynch MCP endpoint to your AI client's configuration. Pass your API key as a Bearer token in the Authorization header.
Start Calling Tools
Your AI client discovers available tools automatically via MCP. Ask it to search contacts, check your pipeline, or create tasks.
Authentication & Transport
Authentication
All requests require a Bearer token: either a Supabase JWT (for the Nynch app) or a scoped MCP API key (for external agents).
- read: search, browse, and retrieve data
- read-write: read plus create, update, delete records
- admin: full access including agent orchestration
Transport
The MCP server uses JSON-RPC 2.0 over Streamable HTTP, the standard MCP transport.
- POST: send tool calls and receive results
- GET /sse: subscribe to real-time events via Server-Sent Events
- Rate limit: 1,000 requests/hour per API key
Tool Catalog
42 tools across 4 categories. Your AI client discovers these automatically via the MCP tools/list method.
CRM Read (10 tools)
Search, browse, and retrieve CRM data
CRM Write (12 tools)
Create, update, and manage CRM records
Relationship Intelligence (11 tools)
AI-powered insights on relationships, deals, and pipeline health
Agent Orchestration (8 tools)
Register AI agents, share context, and hand off tasks between agents
MCP Resources
Resource templates for reading structured context directly.
Full contact context: profile, relationship score, recent interactions, notes, and network.
Deal context: details, stakeholders, coaching actions, learnings, and relationship network.
Pipeline health overview: stage distribution, active value, at-risk deals, and upcoming closes.
Agent Orchestration
Run multiple AI agents that share context and hand off tasks to each other, all through the same MCP endpoint.
How It Works
- Each agent registers with a name, capabilities, and archetype
- Agents share state via a key-value context store with namespaces and TTLs
- Tasks are routed by agent name or capability matching
- Heartbeats track agent liveness; stale agents are flagged
Archetypes
- Producer: creates data (research, enrichment)
- Consumer: reads and acts on data (outreach, reporting)
- Coordinator: orchestrates other agents
- Critic: reviews and validates output
Code Examples
{
"mcpServers": {
"nynch": {
"url": "https://crcrwtjzqzinfzokianr.supabase.co/functions/v1/mcp-server",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
} {
"mcpServers": {
"nynch": {
"url": "https://crcrwtjzqzinfzokianr.supabase.co/functions/v1/mcp-server",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
} from mcp import ClientSession
from mcp.client.streamable_http import streamablehttp_client
async with streamablehttp_client(
"https://crcrwtjzqzinfzokianr.supabase.co/functions/v1/mcp-server",
headers={"Authorization": "Bearer YOUR_API_KEY"}
) as (read, write, _):
async with ClientSession(read, write) as session:
await session.initialize()
# Search contacts
result = await session.call_tool(
"search_contacts", {"query": "Acme Corp"}
)
print(result) Ready to build?
Create a free Nynch account, generate an API key, and start building in minutes.
Get Started Free