Litecode

Litecode

Litecode is a desktop client that helps you plan, run, and review AI-assisted development work across multiple repositories. Manage tasks, orchestrate AI agents, run interactive terminals, and review branches without leaving your workstation.

Download for macOS, Windows & Linux Get started

How Litecode works

Litecode is built around a development workflow that keeps AI agents productive without letting them run unsupervised. Five things make that possible:

  1. 1

    A workspace per repo & branch

    Litecode manages a git worktree for every branch you touch, so each branch gets its own terminals, Claude sessions, Canvas, and notes. Switching contexts is instant and never disturbs work in flight on another branch.

    my-app repo main worktree Terminal Agent session Canvas Tasks feat/login worktree Terminal Agent session Canvas Tasks fix/bug-42 worktree Terminal Agent session Canvas Tasks
  2. 2

    Pull down remote agent branches in one click

    When Claude Code (or any other agent) pushes a branch from a remote session, Litecode lets you check it out, review the diff, and run it locally without juggling git commands or stash dances.

    Claude Code remote session runs in the cloud push origin/feat/abc branch on GitHub one click Litecode local worktree review · run · iterate
  3. 3

    First-class SPDD support

    A dedicated Design tab brings REASONS Canvases into the UI. The Canvas is the source of truth, and Litecode keeps source code in sync with it — generating from the Canvas for behavior changes and syncing the Canvas back from non-behavior refactors.

    /story /analysis /reasons-canvas prompt-update sync SOURCE OF TRUTH REASONS Canvas spdd/prompt/*.md R · E · A · S · O · N · S generate Code · Tests verified by /spdd-api-test
  4. 4

    Request changes — don't make them

    Leave inline comments on any Canvas section or source file, then ask Claude to address them. You drive intent through review, the agent does the editing, and nothing changes without a comment to anchor it.

    Canvas / source file "Use Y instead of X here." — you "address comments" Claude edits Diff for you to review
  5. 5

    First-class Docker integration

    Run coding agents inside Docker with one toggle. You decide which MCP servers and environment variables are exposed to the container, so agents get exactly the capabilities you grant them — and nothing more.

    MCP servers github filesystem …allow-listed Env vars GITHUB_TOKEN PATH …allow-listed DOCKER CONTAINER Coding agent

Capabilities at a glance

Beyond the workflow itself, Litecode is a full desktop workspace for AI-assisted development across one or many repositories.

Task management

Create, organize, and track development tasks in named views, with sub-tasks, plans, and references that link directly to repos and branches.

AI agent orchestration

Configure autonomous agents (Claude Code, Codex CLI, Gemini, Aider, and more) with schedules, inboxes, and per-agent state stores.

Git & GitHub integration

Branch workflows, PRs, and reviews are first-class — including inline review comments on diffs across your tracked repos.

Interactive terminals

Spin up agent-aware terminal sessions anchored to specific repos and branches, with MCP tooling baked in.

Docker dev environments

Run sessions inside reproducible containers so each task gets a clean, isolated environment.

SPDD pipeline integration

Drive Structured Prompt-Driven Development end to end — from story through canvas, generation, and review.

Mobile remote access

Check on tasks, agents, and notifications from your phone via the built-in remote access API.

Code analysis & visualization

Inline Mermaid diagrams, SVG rendering, and a long-form readability mode make documentation and reviews easier on the eyes.

Latest from the blog

Browse all posts →