project
CodeLeveler.
A local-first coding agent runtime that inspects, edits, runs, and verifies real repositories.
What is CodeLeveler?
CodeLeveler is a local-first coding agent runtime. It can enter a real repository, understand a task, edit files, run commands, verify the result, and produce a reviewable diff.
Why it exists
Calling a model is only one part of a coding agent. The harder work is durable sessions, tool execution, permissions, workspace isolation, checkpoints, and recovery. CodeLeveler brings these concerns into one runtime shared by its terminal, Web, and CLI clients.
Current capabilities
- TUI, Web UI, and CLI
- Local sessions, permissions, and project state
- Workspace, approval, and checkpoint controls
- Bring your own model provider
Current status
The project is in public beta and currently centers on the terminal workflow.