A practical way for Pega Solution Designers to prototype with customer AI teams
The value of Blueprint MCP Preview is simple: it enables rapid prototyping workflows with customer AI teams and systems, so you can demonstrate the end-to-end experience faster.
The challenge today is that custom connectors required for the Blueprint MCP Preview are not universally accessible across AI clients for all users (for example, in Claude). Because of that, you may need a local integration approach: Universal MCP Gateway that lets you bring you Blueprinted workflows to life with Customer AI systems.
A Universal MCP Gateway addresses this by providing:
- One executable
- Windows + macOS support
- STDIO + HTTP transport
- Multi-endpoint routing
- Secure token injection
- Local-only operation
MCP Gateway Architecture
Quick Setup (High-Level)
- Install Go: Install Go
- Clone or download the gateway repository: Pega-blueprint-mcp-gateway Github Project
- Create a local config file with your Blueprint MCP preview URL/token.
- Run the gateway in STDIO mode.
- Add the gateway command to your client MCP configuration.
- Restart the AI client.
- Validate that Blueprint MCP tools/resources are available.
For full details, use FIRST_TIME_CLAUDE_SETUP.md.
Claude Desktop Example (macOS)
Use this in your Claude MCP configuration (replace the path with your local repo path):
{
“mcpServers”: {
“mcp-preview-blueprint”: {
“command”: “/bin/zsh”,
“args”: [
“-lc”,
“cd ‘/ABSOLUTE/PATH/TO/pega-blueprint-mcp-gateway’ && go run ./cmd/universal-mcp-gateway -mode stdio -config config.local.json”
]
}
}
}
This setup makes Blueprint MCP Preview usable in real client environments where native connector support is inconsistent—so Solution Designers can focus on design and outcomes, not integration friction.