Amplify MCP Proxy
Exposes tools to query Salesforce for candidates by name and send them to a BHA screening API. Designed for deployment on AWS Lambda with API Gateway.
README
Amplify MCP POC Proxy
A lightweight MCP server exposing two tools against a Salesforce org and BHA API. Deployed as AWS Lambda + API Gateway (HTTP API).
Tools
| Tool | Description |
|---|---|
find_candidate_by_name |
SOQL query against Salesforce Contacts (Candidate record type) by name |
send_candidate_to_screening |
POSTs to BHA API with candidate Id, returns screening URL |
Prerequisites
- Node.js 18+
- AWS CLI configured (
aws configure) - AWS SAM CLI (
brew install aws-sam-clior see docs) - A Salesforce org with a Connected App configured for username/password OAuth
- BHA static API token
Local Setup
# 1. Install dependencies
npm install
# 2. Copy and fill in env vars
cp .env.example .env
# Edit .env with your real credentials
# 3. Run local smoke test against your SF org
node test-local.js "Smith"
The smoke test runs the full MCP handshake locally without deploying anything. If you see candidate names printed, the Salesforce connection is working.
Salesforce Connected App Setup
In your trial org (Setup > App Manager > New Connected App):
- Enable OAuth Settings
- Callback URL:
https://login.salesforce.com/services/oauth2/success(not used but required) - Selected OAuth Scopes:
- Access and manage your data (api)
- Perform requests at any time (refresh_token, offline_access)
- Enable Client Credentials Flow: checked
- Save -- note the Consumer Key (Client ID) and Consumer Secret
Also ensure the running user has API access and can read Contact records.
Deploy to AWS
# 1. Build the deployment package
cd infra
sam build --template template.yaml --build-dir .aws-sam
# 2. Deploy (guided first time -- saves config to samconfig.toml)
sam deploy --guided \
--template template.yaml \
--stack-name amplify-mcp-poc \
--capabilities CAPABILITY_IAM \
--parameter-overrides \
SfLoginUrl=https://test.salesforce.com \
SfClientId=YOUR_CLIENT_ID \
SfClientSecret=YOUR_CLIENT_SECRET \
SfUsername=your.user@org.sandbox \
SfPassword=yourpassword \
SfSecurityToken=yourtoken \
BhaApiBaseUrl=https://api.herefish.com/api/v1 \
BhaApiToken=YOUR_BHA_TOKEN
# 3. Note the McpEndpoint output URL -- you'll need this for Claude Desktop
Subsequent deploys (after the first):
sam deploy --template template.yaml --stack-name amplify-mcp-poc
Wire into Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"amplify-poc": {
"url": "https://YOUR_API_ID.execute-api.eu-west-1.amazonaws.com/mcp",
"transport": "http"
}
}
}
Replace YOUR_API_ID with the value from the McpEndpoint SAM output.
Restart Claude Desktop -- you should see find_candidate_by_name and
send_candidate_to_screening appear as available tools.
Project Structure
amplify-mcp-proxy/
├── src/
│ ├── handler.js # Lambda entry point -- MCP protocol router
│ ├── tools.js # Tool definitions + handlers
│ ├── salesforce.js # SF OAuth + SOQL query helper
│ └── bha.js # BHA API client
├── infra/
│ └── template.yaml # AWS SAM template
├── test-local.js # Local smoke test
├── .env.example # Environment variable template
└── package.json
Extending
To add a new tool:
- Add the definition + handler function in
src/tools.js - Register it in the
TOOLSmap at the bottom of that file - That's it --
tools/listandtools/callrouting pick it up automatically
Notes
-
The MCP session is stateless -- no server-side session storage. A fresh
mcp-session-idis issued on everyinitializecall but no state is stored. This matches the production Bullhorn proxy pattern. -
The Salesforce access token is cached in Lambda memory for the lifetime of the execution context (warm invocations reuse it). Cold starts re-authenticate. This means one auth call per cold start, not per tool call.
-
The BHA endpoint path (
/screening) insrc/bha.jsmay need adjusting to match your actual BHA API route. Check with your BHA/Herefish contact for the exact endpoint and payload shape.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.