OpenCode HopGPT Bridge

A complete pipeline that connects OpenCode (AI-powered IDE) to JHU’s HopGPT service, turning a web-only university AI chat interface into a fully programmable development environment. The system spans browser automation, API translation, token management, and IDE configuration.

The Stack

┌─────────────┐      ┌─────────────────┐      ┌──────────────────┐      ┌─────────────┐
│   OpenCode  │─────►│  FastAPI Server │─────►│  Browser Service │─────►│   HopGPT    │
│    (IDE)    │ HTTP │  (Port 8000)    │  ZMQ │  (Playwright)    │ CDP  │ (JHU's AI)  │
└─────────────┘      └─────────────────┘      └──────────────────┘      └─────────────┘

Components

1. Browser Service (Playwright Daemon)

• Persistent Chromium instance with stealth settings to avoid automation detection
• Automated Microsoft SSO login flow (email → password → MFA handling → session persistence)
• Chrome DevTools Protocol (CDP) network interception for real-time token capture
• ZeroMQ IPC server for communication with the API layer
• Automatic session recovery when authentication expires

2. FastAPI API Server

• OpenAI-compatible /v1/chat/completions endpoint
• Request translation from OpenAI format to HopGPT’s internal API structure
• Automatic model routing based on model family:
  • Claude models → /api/agents/chat/AnthropicClaude
  • GPT models → /api/agents/chat/AzureOpenAI
  • Llama models → /api/agents/chat/MetaLlama
• SSE streaming response parsing and reformatting
• Retry logic with automatic token refresh on 401 errors

3. Token Management System

• CDP’s Network.requestWillBeSent event captures Authorization headers passively
• No cookie parsing or localStorage manipulation needed
• Detects null/invalid tokens and triggers refresh automatically
• Token refresh via UI automation: navigates to new chat, sends test message, captures fresh token from the authenticated request

4. OpenCode Configuration

• Custom provider configuration pointing to localhost:8000
• Model definitions matching HopGPT’s available models (Claude Opus 4.5, Claude Sonnet 4.5, GPT-4o, Llama 4, etc.)
• Seamless integration - OpenCode treats it as any other OpenAI-compatible provider

5. Custom Prompt System

• JSON-based configuration for system prompts and personas
• Hot-reloadable without server restart
• Supports multiple named configurations with easy switching

Setup & Infrastructure

The project includes:

• Makefile for service orchestration (make start, make stop, make status)
• Environment configuration for credentials and model defaults
• PostgreSQL integration for conversation persistence (optional)
• Logging infrastructure for debugging token flows and API translation

Technical Highlights

• Zero API Keys Required: Uses existing university authentication - if you can log into HopGPT in a browser, this works
• Transparent Proxying: OpenCode (or any OpenAI-compatible tool) has no idea it’s talking to a browser automation layer
• Graceful Degradation: Token expiration, network issues, and session timeouts are handled automatically with retry logic
• Model Agnostic: Same interface works for Claude, GPT, and Llama models - just change the model name in your request

Outcome

Full IDE integration with university-provided AI models. Write code with Claude Opus 4.5 assistance, get explanations, refactor - all through OpenCode’s native interface, all running through JHU infrastructure. No API keys, no usage limits beyond what HopGPT allows, no cost.

Repo Link

Private (university credentials required)