What is heapcode?

Heap Code is a coding assistant built for VS Code that works with almost any AI model provider, whether that is a cloud service like OpenAI, Azure, OpenRouter, Groq, or Together AI, or a model you run yourself with tools like Ollama, LM Studio, vLLM, or LocalAI. Instead of locking you into one company's AI, it lets you set up named profiles for different providers and switch between them from the status bar, choosing separate models for chat, code editing, autocompletion, the autonomous agent, and search. The tool offers a chat panel with syntax highlighted code and slash commands such as /explain, /fix, and /review, plus the ability to reference your current file, selection, or the whole workspace in a question. It also provides ghost text completions as you type, an inline edit mode where you select code and describe a change, and an agent mode that can read, search, and edit files or run commands on its own. Every non read action asks for permission first, and a one click Revert all option restores every file the agent touched. For understanding a codebase, Heap Code builds a semantic search index using an embeddings model, combining it with plain keyword search so it can find relevant code even without a heavy AI model behind it. It supports the Model Context Protocol so extra tools from other servers can be added, and it reads project notes and instructions the same way Claude Code does, so any memory files already written for other AI tools can be reused. Because your code only leaves your machine to reach whichever model endpoint you configure, it is possible to run entirely offline with local models like Ollama and keep everything private, though the extension does collect anonymous usage telemetry by default that can be turned off in settings. The project is licensed under PolyForm Noncommercial 1.0.0, which means it is free for personal, research, and internal use but cannot be sold or offered as a paid hosted service. Earlier versions before 0.2.0 remain available under Apache-2.0.