Skip to main content
Glama
Sign In
Pinpole MCP Server

Pinpole MCP Server

by codeforstartups
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Remote

Pinpole MCP Server

Design cloud architectures, run cost/performance simulations, and draw them on your Pinpole canvas — from Cursor, Claude Code, Codex, Claude.ai, and Bolt.

Product: https://pinpole.cloud · App: https://app.pinpole.cloud · Connectors: docs/connectors.md

Quick start (stdio — Cursor, Claude Code, Codex)

1. Get a token

Open Pinpole → Settings → Developer / MCP and create a personal access token (pp_live_…). Copy it — it's shown only once.

2. Configure your agent

Cursor — add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "pinpole": {
      "command": "npx",
      "args": ["-y", "@pinpole/mcp"],
      "env": {
        "PINPOLE_API_TOKEN": "pp_live_…",
        "PINPOLE_BASE_URL": "https://app.pinpole.cloud"
      }
    }
  }
}

Claude Code:

claude mcp add pinpole \
  --env PINPOLE_API_TOKEN=pp_live_… \
  --env PINPOLE_BASE_URL=https://app.pinpole.cloud \
  -- npx -y @pinpole/mcp

Codex — add to ~/.codex/config.toml:

[mcp_servers.pinpole]
command = "npx"
args = ["-y", "@pinpole/mcp"]
env = { PINPOLE_API_TOKEN = "pp_live_…", PINPOLE_BASE_URL = "https://app.pinpole.cloud" }

Related MCP server: Synlake MCP Server

Remote connector (Claude Desktop, Claude.ai, Bolt)

Field

Value

URL

https://app.pinpole.cloud/mcp

Transport

HTTP

Auth

OAuth (Claude) or API key pp_live_… (Bolt)

Claude Desktop: Settings → Connectors → Add custom connector → URL above (leave OAuth Client ID/Secret blank). See docs/connectors.md.

OAuth discovery: https://app.pinpole.cloud/.well-known/oauth-authorization-server

Self-hosted HTTP server:

npm run build && npm run start:http

Env: MCP_HOST, MCP_PORT (default 3333), PINPOLE_BASE_URL.

Embed in your Node app:

import { createMcpApp } from "@pinpole/mcp/http";
const app = createMcpApp();
app.listen(3333);

Tools

66 MCP tools covering projects, workspaces, AI architect, simulation, deploy, templates, teams, connectors, billing, and export. Full API parity matrix: docs/PARITY.md.

Highlights:

Tool

What it does

pinpole_build_architecture

Prompt → architecture → inline canvas in chat (+ optional simulation)

pinpole_ai_chat / pinpole_ai_vision

Multi-turn AI architect

pinpole_simulate_cost

Cost/latency simulation

pinpole_list_projects / workspace CRUD

Project & canvas management

pinpole_open_canvas

Inline read-only canvas embed (PAT) + links to full editor

Canvas in Cursor

With PINPOLE_API_TOKEN set, canvas tools mint a short-lived /embed/canvas URL and return an inline read-only diagram in chat (no Pinpole browser login). Use Open interactive canvas in Pinpole for full drag-and-drop editing.

See docs/connectors.md. | pinpole_deploy_execute / pinpole_get_drift | Deploy & drift | | pinpole_export_terraform | Offline Terraform export (no network) |

Environment variables

Variable

Default

Notes

PINPOLE_API_TOKEN

–

pp_live_… from Developer settings

PINPOLE_BASE_URL

https://app.pinpole.cloud

API base URL

PINPOLE_DEV_USER_ID

–

Local dev only (ALLOW_DEV_USER_HEADER=1 on server)

Development

npm ci
npm run build
npm test
npm test              # smoke: tools/list
npm run test:tools    # list all registered tools
npm run test:parity   # compare tools against docs/PARITY.md

Local integration testing (no browser UI): see docs/LOCAL_TESTING.md.

Links

MIT licensed.

Available Tools

9 tools
pinpole_build_architectureBuild architecture (design + draw on canvas)A

Turn a natural-language system description into a validated cloud architecture and draw it on the user's Pinpole canvas. Returns a canvas URL the user can open to see and edit the live ReactFlow diagram (a draw.io replacement). Optionally runs a cost simulation. Use this when the user wants to design or visualize a system/architecture.

Inputs: prompt (required), title, projectId, workspaceId, +1 more.

pinpole_create_architectureCreate architecture graph from a promptA

Generate a validated {nodes, edges} cloud architecture from a description using Pinpole's AWS Well-Architected model. Does NOT draw it — pass the result to pinpole_draw_on_canvas or pinpole_simulate_cost. Use pinpole_build_architecture to do both in one step.

Inputs: prompt (required), existingNodes.

pinpole_create_projectCreate a Pinpole projectC

Create a new Pinpole project to hold architecture workspaces.

Inputs: title (required), description.

pinpole_draw_on_canvasDraw nodes + connections on the Pinpole canvasA

Persist an architecture graph to the user's Pinpole canvas (ReactFlow), creating a project and workspace if none is given. Returns the canvas URL to open the live diagram. Nodes are auto-laid-out left-to-right when positions are omitted.

Inputs: graph (required), title, projectId, workspaceId.

pinpole_export_terraformExport Terraform from an architectureC

Generate Terraform project files from an architecture graph. Works offline.

Inputs: graph (required).

pinpole_list_projectsList Pinpole projectsA

List the authenticated user's Pinpole projects (id + title).

Inputs: no named parameters.

pinpole_list_servicesList available AWS service specsA

List AWS service ids/specs the canvas understands. Reads local repo specs when available.

Inputs: no named parameters.

pinpole_open_canvasGet the canvas URL for a workspaceB

Return the Pinpole canvas deep link for a project + workspace so the user can open the live, editable architecture diagram in their browser.

Inputs: projectId (required), workspaceId (required).

pinpole_simulate_costRun a traffic simulation and get costA

Run Pinpole's simulation engine over an architecture graph at a given traffic level and return per-node + total estimated monthly cost, latency (p50/p95/p99), throughput, and throttling/error alerts. Use to answer 'what will this cost / how will it perform'.

Inputs: graph (required), traffic (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Deploy Server
    One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

    TDQS

    A3.8/5.0

    Scored across 9 tools

    Disambiguation5/5

    Each tool has a clearly distinct purpose: building vs creating architecture, drawing vs simulating, listing vs opening. No overlapping functionality that could confuse an agent.

    Naming Consistency5/5

    All tools follow a consistent 'pinpole_verb_noun' pattern in snake_case, making the naming predictable and easy to understand.

    Tool Count5/5

    With 9 tools, the server covers the main workflows (create, visualize, simulate, export) without being bloated or too sparse. The count feels well-scoped.

    Completeness4/5

    The core lifecycle of architecture design is covered: create project, define architecture, draw on canvas, simulate cost, export to Terraform. Minor gaps like update/delete operations are absent but can be handled via the canvas UI.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Resources

    Related MCP Connectors

    Related MCP Servers