peterjthomson/ledger

A modernised git interface for improved agent and human collaboration, review and control

What it solves

Ledger is a modernised git interface designed to handle the high volume and velocity of code changes generated by AI coding agents and modern software teams. It provides a "command & control radar" for developers to oversee changes across multiple local and remote agents, reducing the friction of managing complex version control workflows.

How it works

Ledger acts as a local desktop application (built with Electron and React) that shells out to git and the GitHub CLI (gh) to perform operations. It provides a visual interface for managing pull requests, worktrees, branches, and stashes, with specific features to detect AI agent workspaces (such as those from Cursor or Claude) and automate common git operations like auto-stashing.

Who it’s for

Software developers who use AI coding agents and need a more efficient way to manage multiple concurrent workstreams, branches, and pull requests on macOS (with support for Linux and Windows).

Highlights

  • AI Agent Workspace Detection: Specifically detects and manages workspaces used by AI agents.
  • Pull Request Integration: View and create GitHub PRs, including inline diffs and reviews, via the gh CLI.
  • Advanced Worktree Support: Navigate and convert worktree changes to branches with change stats per worktree.
  • Granular Staging: Stage or unstage specific files, hunks, or individual lines with character-level diff previews.
  • Opinionated Git Behaviors: Includes "virtual merge" PR previews and auto-stashing for safe operations.

Related

  • Project
  • Project
  • Project
  • Project