AI-powered multi-platform messaging agent that runs inside VS Code. Ask anything in natural language and Phoenix decides the right action — answering, searching the web, or preparing a Telegram / Gmail draft for you to review and send.
Features
Multi-provider LLM support — NVIDIA, OpenAI, and Groq backends with a per-chat model picker.
Smart step routing — queries are classified and routed to the right tool:
general — direct AI answers
web_search — real-time web results (Tavily / Serper)
telegram — draft a Telegram message for your chat IDs
gmail — draft an email for your recipients
Draft & review flow — AI writes the message, you edit and approve before it is sent.
Full RTL support — correct editing and rendering for Persian, Arabic, Hebrew, and other RTL languages.
Session persistence — chat history stored in MongoDB (sessions + messages) with a left sidebar, rename-on-first-message, and per-session delete.
Dark, chat-style UI with a sidebar, session list, and a brand footer.
Requirements
VS Code 1.85.0 or newer
MongoDB running locally (default mongodb://127.0.0.1:27017) or any MongoDB URI you set
API keys for the services you want to use (see Settings below)
Usage
Open the command palette (Ctrl+Shift+P) and run Phoenix: Open Agent Panel.
Configure your API keys and MongoDB URI via Settings → Extensions → Phoenix Agent, or click the ⚙️ button in the panel.
