Remote AI Coding assist VS Code

# Kilo Code: Long Description ## The Problem As agentic engineering becomes standard, developers are tethered to their workstations to monitor long-running AI coding tasks and provide critical permissions. This bottleneck reduces productivity and prevents real-time responsiveness during urgent development cycles. ## The Solution: Remote Agentic Management Kilo Code transforms development by providing a secure Telegram-based management layer for your AI agents. You can monitor progress, authorize sensitive operations (like shell commands or file writes), and interact with your agent via natural language from any location. ## AMD Integration & Qwen3.6 At the heart of Kilo Code is the power of the AMD Developer Cloud. We leverage Qwen3.6, a state-of-the-art model with exceptional performance on complex coding tasks, reasoning, and tool-use capabilities. By utilizing the AMD MI300X GPU stack with ROCm, Kilo Code ensures ultra-low-latency inference, enabling complex agentic loops to run with remarkable speed. ## Flexibility with Local LLMs We prioritize developer flexibility. While optimized for AMD Cloud, Kilo Code natively supports local LLM execution via Ollama. Developers can maintain maximum privacy by keeping their code context local, while still leveraging the high-speed coding prowess of Qwen3.6 on their local AMD hardware or remote infrastructure. ## Architecture - **Kilo Core:** Manages the agentic lifecycle and planning. - **Telegram Interface:** Provides mobile-friendly remote orchestration, event notifications (SSE), and rich visualization (file trees, diffs). - **Inference Layer:** Uses Qwen3.6 via AMD ROCm/MI300X or locally through Ollama, delivering superior tool-use and reasoning performance. - **Security:** Integrated VS Code SecretStorage and chat-ID whitelisting to ensure secure remote interaction.