💰 FUNDING NEWS: Hushh.ai Secures $5 Million Strategic Investment from hushhTech.com's Evergreen Renaissance AI Fund
💰 FUNDING NEWS: Hushh.ai Secures $5 Million Strategic Investment from hushhTech.com's Evergreen Renaissance AI Fund
💰 FUNDING NEWS: Hushh.ai Secures $5 Million Strategic Investment from hushhTech.com's Evergreen Renaissance AI Fund
Hushh Plaid Financial Profile Creation Agent
Create and sync Plaid financial profiles with MCP + Supabase in one flow.
This page wraps the end-to-end JSON-RPC flow described in the platform docs: capture a user instruction, let the agent extract structured fields, and push a new financial profile into Supabase through the MCP server.
LLM model
OpenAI GPT-4.0 mini
Transport
JSON-RPC 2.0 via MuleSoft
Data store
Supabase via MCP insert_user
Overview
An AI-driven automation agent that converts natural language or semi-structured financial inputs into new Supabase user profiles. It keeps personalization, analytics, and recommendations in sync with Plaid-sourced data.
Account + contact enrichment
Transactions, investments, liabilities
Timestamped confirmations
Integration Flow
1) MuleSoft crafts JSON-RPC from a user or backend instruction.
2) Agent parses fields and calls the MCP server.
3) MCP insert_user writes to Supabase.
4) Structured confirmation returns to the Hushh interface.
Functional Workflow
Step 1
User input
Natural language prompt (or backend trigger) containing raw financial profile details.
Step 2
Agent invocation
MuleSoft crafts a JSON-RPC 2.0 POST to the Plaid Financial Profile Creation Agent.
Step 3
Intent extraction
LLM parses account, contact, transactions, investments, and liability data points.
Step 4
MCP + Supabase
MCP server executes insert_user to persist the unified record in Supabase.
Step 5
Confirmation
Agent returns a structured success payload with timestamps and tracking IDs.
Quick testing presets
Load sample values from the documentation to test the flow in one click.
Agent Request Builder
Compose the JSON-RPC request the same way MuleSoft does. The agent consumes your natural language prompt plus the embedded profile JSON and then calls MCP for insertion.
Session ID
Task ID
User prompt
Financial profile payload
Request preview
Add or load a profile JSON payload to preview the request.Direct MCP call (insert_user)
Validate how MCP ingests structured arguments without the agent layer. Use the same request object your MuleSoft flow would POST.
tools/call payload
Response console
Responses from the agent and MCP are captured here for quick debugging and demo purposes.
No calls yet. Trigger a request to see responses.
Error handling
400
Invalid JSON-RPC format
Validate request schema and required keys.
401
Unauthorized
Check MuleSoft credentials and headers.
404
Required fields missing
Provide mandatory profile parameters.
409
Duplicate user
Verify user identity before retrying.
500
Internal Agent Error
Retry or escalate to the platform team.
TIMEOUT
MCP server delay
Increase timeout or retry the request.
Security & privacy
• HTTPS enforced across agent and MCP endpoints.
• Only authorized MuleSoft flows can initiate profile creation.
• Supabase credentials remain isolated inside MCP.
Versioning
v1.0.0 — Initial Hushh Plaid Financial Profile Creation Agent release.
API endpoints
Agent: https://hushh-plaid-agent-app-bubqpu.5sc6y6-4.usa-e2.cloudhub.io/plaid-agent
MCP Server: https://hushh-plaid-mcp-server-app-bubqpu.5sc6y6-4.usa-e2.cloudhub.io/mcp
Method: POST with Content-Type: application/json