Skip to content

Keystone Lite v1.0.0 — Local-First AI Vibe Code Editor

Choose a tag to compare

@ethgr0wth ethgr0wth released this 18 Jan 03:17
· 98 commits to main since this release

Keystone Lite v1.0.0

This release introduces Keystone Lite, a local-first desktop environment for AI-assisted vibe code editing and debugging.

The goal is simple: enable precise, auditable interaction between developers and AI without surrendering control of the codebase. All edits are explicit, line-based, and reviewable. Automation is optional.

Keystone Lite operates entirely on the user’s machine. Model choice is external and user-supplied. No telemetry is collected. File access is scoped and intentional. Agentic workflows have made a significant difference in my life, I enjoy vibe coding. Hope that you fine this useful.

This version includes a Monaco-based editor, conversational file context, dual execution modes (review and auto-apply), and a set of starter project templates to accelerate development.

The software is released as source-available under the Business Source License 1.1, transitioning to MIT in the future. The intent is to encourage experimentation, forks for learning, and transparent iteration while preserving a sustainable path forward.

This is the first public release.

Interface Overview

Keystone Lite
Keystone Lite
Keystone Lite

Features

  • Local-first desktop application for AI-assisted code editing
  • Monaco-based editor with full language support
  • Conversational interface with active file context and line numbers
  • Deterministic, line-based edit format (INSERT / REPLACE / DELETE)
  • Dual execution modes: review-first (Debug) and automatic apply (Keystone)
  • Bring-Your-Own-AI support for 11 compatible providers
  • File system access scoped to the opened project directory
  • Encrypted local storage for API credentials
  • No telemetry, tracking, or background network activity
  • Built-in project templates for rapid initialization
  • Cross-platform architecture (Windows release, others forthcoming)
  • Streaming support with client-side buffer utility