Jackalope

Product Introduction

1. Definition: Jackalope is a cross-platform, open-source desktop application that functions as a unified orchestration workspace for multiple AI-powered coding agents. It technically integrates with agent CLIs (Command Line Interfaces) like Codex, Claude Code, Grok, OpenCode, Kimi Code, and Antigravity, managing their execution within isolated Git worktrees.
2. Core Value Proposition: Jackalope exists to eliminate context-switching and manual coordination overhead when using multiple AI coding assistants. Its primary value is enabling parallel AI coding agent execution within a single, context-aware workspace, allowing developers to run simultaneous tasks, maintain persistent project knowledge, and conduct consolidated code reviews before merging changes.

Main Features

1. Parallel Agent Execution in Git Worktrees: Jackalope launches each AI agent task in a separate, isolated Git worktree. This provides a clean, concurrent sandbox for each agent (e.g., Codex building a feature while Claude Code fixes a bug). Changes remain separate until explicitly reviewed and combined, preventing task interference and preserving the main branch.
2. Persistent Project Context & MCP Integration: The workspace stores reusable project instructions, guidelines, and tool configurations. This context is automatically matched to relevant new tasks. It integrates with the Model Context Protocol (MCP) for connecting external tools and data sources, providing agents with a shared, enriched knowledge base without repetitive manual briefing.
3. Unified Code Review with Dependency Visualization: All agent-generated changes are presented for review in a single interface. Jackalope provides an interactive codebase visualization map that visually traces file dependencies and the potential impact of changes, helping developers assess modifications before applying them. Related patches from different tasks can be combined for holistic review.
4. Multi-Account Management & Usage Tracking: It supports named sign-in profiles for different provider accounts (e.g., work vs. personal), allowing project-level defaults. Jackalope tracks reported usage and quotas per agent and account, facilitating cost management and ensuring tasks use the correct credentials.
5. Built-in Agent Tooling (Browser & Computer Use): Agents are equipped with a built-in, isolated browser for web navigation, form filling, and accessibility audits. On Windows, agents can be granted controlled access to a specific application window for clicks, typing, and scrolling, with a clear visual indicator and user-override controls (Escape to cancel).
6. Recurring & Automated Tasks: Tasks can be scheduled to run automatically or triggered by changes in a local Git branch. Change detection operates without model calls, making it efficient for automated linting, testing, or dependency updates. Each run's history and results are preserved within the workspace.

Problems Solved

1. Pain Point: Developers manually juggle multiple terminal tabs, copy-pasting context between different AI agent sessions, leading to lost information, repetitive briefing, and an inability to manage concurrent explorations or fixes efficiently.
2. Target Audience: The primary user personas are Full-Stack Developers, Engineering Managers, and Open-Source Maintainers who regularly use AI coding assistants like Claude Code or Codex for feature development, bug investigation, code refactoring, and automated codebase maintenance.
3. Use Cases: Essential scenarios include: Parallel prototyping (testing two different AI-generated implementations side-by-side); Stubborn bug investigation (assigning different agents to explore separate hypotheses); Scheduled codebase hygiene (running automated dependency updates or linting with an agent); and Consolidated refactoring (orchestrating multiple agents to update interconnected files with a unified review).

Unique Advantages

1. Differentiation: Unlike using standalone agent CLIs or IDE plugins, Jackalope is not tied to a single agent. It is a meta-orchestrator that manages the lifecycle, context, and output of multiple, heterogeneous AI coding agents from different providers within a Git-native workflow, which is a unique approach in the market.
2. Key Innovation: The core innovation is the context-aware, worktree-isolated task model. By binding each agent task to a Git worktree and a durable project context, it creates a reproducible, reviewable unit of AI-assisted work. This, combined with cross-agent communication channels (direct messages, project broadcasts) and quota-aware routing, enables a coordinated multi-agent system rather than a collection of isolated tools.

Frequently Asked Questions (FAQ)

1. Does Jackalope require an AI subscription? No, Jackalope itself does not provide or include model access. You must bring your own supported AI agent CLIs (e.g., Claude Code, Codex) and their respective provider accounts/subscriptions. Jackalope orchestrates their use.
2. Is my source code sent to Jackalope's servers? No, all repositories, Git worktrees, and task workspaces reside locally on your computer. Code and context are only sent to AI providers according to the settings of the individual agent CLIs you have configured and authorized.
3. How does Jackalope handle work and personal AI accounts? You can create named account profiles (e.g., "Work OpenAI," "Personal Anthropic") within Jackalope. Each project can be configured with a default allowed account, keeping usage and billing separate. Note: profiles manage sign-ins but do not sandbox local file system access.
4. Can Jackalope automatically choose which AI agent to use for a task? Yes, for "automatic" tasks, your configured default agent can perform quota-aware routing. It selects from your project's allowed lineup of agents and accounts based on reported capacity and can hand off work if a quota limit is hit, preserving the task's context and progress.
5. What operating systems does Jackalope support? Jackalope is a cross-platform desktop application. The built-in browser automation works across platforms. The specific "computer use" feature for controlling desktop apps is currently available for Windows, with native validation for macOS and Linux (X11) in development and Wayland not yet supported.