PahamConnect

Persistent context and specification platform for AI coding agents.

PahamConnect acts as the persistent context layer that helps AI agents understand what the developer is working on, why it is being done, what has happened, what decisions were made, and what should happen next.

It bridges the gap between structured development workflows (like OpenSpec) and minimal developer tooling (like DBHub).

Installation

npm install -g @mazkama/pahamconnect

Transport Modes: stdio vs http

PahamConnect supports two connection protocols depending on your use case:

| Mode | Command | Best For | Do you need to run serve? | |---|---|---|---| | stdio (Default) | Spawned by AI client | Local coding assistant (Claude Desktop, Cursor, local IDE) | No (Headless mode, AI handles the process) | | http | pahamconnect serve | Web Workbench dashboard, remote APIs, cloud IDE integrations | Yes (Runs server on port 8787) |

Quick Start (For Web Workbench)

# 1. Initialize DB
pahamconnect init

# 2. Run the HTTP server & Workbench
pahamconnect serve --port 8787

Open your browser to:

• Workbench Dashboard: http://localhost:8787
• MCP SSE Server Endpoint: http://localhost:8787/mcp

Documentation

Full documentation is available in the docs/ folder:

• Landing Page
• Quickstart Guide

Development

# Install dependencies
npm install

# Build TypeScript
npm run build

# Run unit tests
npm test