v.5.0
A lightweight web UI for chatting with multiple AI providers. No frameworks, no build steps — just HTML, CSS, and JavaScript.
- Copy
config.example.json→config.jsonand add your API keys. - Open
index.htmlin a browser.
Tip: For local file:// usage, copy
config.local.example.js→config.local.jsinstead.
| Provider | Models | Auth |
|---|---|---|
| OpenAI | GPT-4o, o1, o3-mini, GPT-5-mini | OpenAI API Key |
| GitHub Copilot (PAT) | GPT-4o, GPT-4o Mini, o3-mini | GitHub PAT |
| GitHub Copilot (ACP) | Claude, GPT-5.x, Gemini 3 Pro | Copilot CLI |
| Google Gemini | Gemini 2.0 Flash Thinking | Gemini API Key |
| lm-studio | Any local model | None |
| DALL-E 3 | Image generation | OpenAI API Key |
- Settings panel with tabbed UI (General, Models, Auth, Prompts, MCP)
- MCP tools — Query Azure Data Explorer (Kusto), GitHub repos, Azure services
- Inline images — Wikimedia search or DALL-E generation
- LCARS theme — Star Trek-inspired interface
- Text-to-speech — Amazon Polly and Bark TTS
- Conversation memory in localStorage
python3 tools/acp_bridge.py --port 8888
python3 tools/acp_bridge.py --port 8888 --enable-kusto-mcp --kusto-cluster "https://..."Requires: copilot CLI installed + authenticated. Details →
See README-2.md for architecture, dependencies, and technical details.
See .github/copilot-instructions.md for contribution guidance.
Based on CodeProject. Complete overhaul.
