Melty
Open-source chat-first AI code editor — watches your terminal, compiler, and git to code alongside you.
About
Melty is an open-source AI code editor built by Charlie Holtz (ex-Replicate) and Jackson de Campos (ex-Netflix) in a 28-day sprint. Instead of inline autocomplete, it centers a persistent chat pane that drives multi-file edits: it watches your terminal output, compiler errors, debugger, and GitHub activity so proposed changes and fixes carry real context about what you are doing. It navigates large codebases, drafts commit messages from diffs, and commits code with you rather than just suggesting snippets.
The honest picture: it is a beta-stage project — at various points only source was available with a waitlist for packaged builds, and development pace has been uneven. The chat-first model is opinionated; developers who live on inline completion may find it slower than Cursor-style flows. Still, the environment-awareness idea was genuinely ahead of its time.
What it can do
- Persistent chat pane driving multi-file edits
- Terminal, compiler, and debugger awareness
- Automatic AI-generated commit messages
- GitHub integration for large-codebase navigation
- Open-source VS Code fork
Setup
Build from source on GitHub (meltylabs/melty) or join the beta for a packaged download. It is a VS Code fork, so settings and extensions carry over.
Prerequisites
An LLM API key for the model backend; familiarity with VS Code workflows.
Use cases
Works with
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 57; Trending score 62/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.