llms.txt
# Jackalope
> A desktop workspace for coding agents, local Git projects, tasks, worktrees, and review.
Jackalope is a product of Jackalope Digital LLC (https://jackalope.digital). The canonical product website is https://jackalope.dev.
## Availability
Desktop downloads are available for Windows through Microsoft Store, macOS, and Linux. See https://jackalope.dev/download/ for current availability. Using the app requires waitlist approval or a claimed Instant Access Pass; model access is separate. Users bring their own locally installed agents and provider accounts; an AI subscription is not included.
## Product
Tasks keep ideas, attempts, results, and review together. Isolated Git worktrees separate working directories. Users inspect patches and run project checks before deciding what to integrate. Website screenshots and the recorded tour use Atlas sample project data.
## Agent support
Native task adapters: Codex, Claude Code, Grok Build, OpenCode, Kimi Code, Gemini CLI, and Antigravity. Direct project MCP connections support Codex, Claude Code, OpenCode, and Kimi Code; all seven adapters support on-demand discovery. Kimi supports named accounts, model selection, structured questions, task tokens, membership quota, routing, and Ask Jackalope. Antigravity is worker-only; its current model and subscription quota use read-only CLI commands. Named profiles use Gemini API keys with separate billing, while subscription login is shared. Gemini CLI runs worker tasks with session continuation and file-edit approval; shell and HTTP tools follow its own permission policy. Its quota is unknown, and it does not coordinate routing or Ask Jackalope. Installed-provider acceptance remains open. Provider access and reported usage vary by adapter. See https://jackalope.dev/agents/ for coverage and limits and https://jackalope.dev/agents/kimi-code/ for Kimi setup.
## Knowledgebase & Diagnostics
- [Agent-readable documentation index](https://jackalope.dev/knowledge/llms.txt)
- [Ask Jackalope and local MCP tools](https://jackalope.dev/knowledge/ask-jackalope/index.md)
Official documentation, architecture guides, and troubleshooting recipes are available at https://jackalope.dev/knowledge/:
- Isolated Git Worktree Architecture: Preventing checkout collisions across parallel agents
- Task Routing & Quota Handoff: Reported quota windows, preflight headroom estimates, and up to three handoffs
- Multi-Account Profiles: Segregating Work and Personal agent provider sign-ins
- MCP Tools & Built-In Browser Automation: Central Model Context Protocol management
- Diagnostic Playbook: Resolving missing CLI PATH, expired tokens, and worktree lock errors
## Pages
- [Jackalope: Run AI coding agents in parallel](https://jackalope.dev/): Run Codex, Claude Code, Gemini CLI and more in parallel, each in its own Git worktree. Review every change in one desktop app for Windows, macOS and Linux.
- [Download Jackalope for Windows, macOS & Linux](https://jackalope.dev/download/): Find Jackalope desktop downloads and platform availability. Sign up free and confirm your email to start.
- [Agent performance benchmarks & methodology | Jackalope](https://jackalope.dev/benchmarks/): Measured task performance through Jackalope and the native agent CLI, with token usage, independent checks, limitations, and reproducible methodology.
- [Jackalope app tour: Tasks, coding agents & code review](https://jackalope.dev/tour/): Explore themes, project setup, task dispatch, browser tools, agent accounts, and the interactive codebase map in a 64-second tour with sample data.
- [Coding-agent workflows & field notes | Jackalope](https://jackalope.dev/blog/): Notes from the Jackalope studio on building a calmer workspace and working thoughtfully with coding agents.
- [Changelog | Jackalope](https://jackalope.dev/changelog/): Follow new Jackalope features, meaningful improvements, and major milestones.
- [Coding-agent guides & troubleshooting | Jackalope](https://jackalope.dev/knowledge/): Explore official guides, parallel worktree architecture, multi-account setup, automatic quota handoff, and troubleshooting recipes for Jackalope.
- [Privacy policy | Jackalope](https://jackalope.dev/privacy/): How Jackalope handles website, early-access, desktop, feedback, and optional usage information.
- [Terms of service | Jackalope](https://jackalope.dev/terms/): Terms for Jackalope’s website and hosted early-access service, with desktop open-source license information.
- [Jackalope roadmap | What comes next](https://jackalope.dev/roadmap/): Follow plans for Jackalope desktop releases, coding-agent support, project tools, and remote work.
- [Compare coding-agent workspaces | Jackalope](https://jackalope.dev/compare/): Compare Jackalope with Codex app, Cursor, Superset, Orca, and other coding-agent workspaces by agent choice, execution, and review.
- [Coding agents: Jackalope or terminal tabs? | Jackalope](https://jackalope.dev/compare/terminal-tabs/): When do terminal sessions work well, and when does a coding-agent workspace help? Compare task context, Git worktrees, questions, and review.
- [Browser automation for coding agents | Jackalope](https://jackalope.dev/features/browser-automation-for-coding-agents/): Keep browser sessions, screenshots, and accessibility findings attached to coding-agent tasks in Jackalope.
- [Jackalope vs Codex app: Workflow comparison](https://jackalope.dev/compare/codex-app/): Jackalope vs Codex app: compare local worktrees, cloud execution, code review, and using Codex alongside other coding-agent CLIs.
- [Jackalope vs Cursor: Workflow comparison](https://jackalope.dev/compare/cursor/): Jackalope vs Cursor: compare agent worktrees, local and cloud execution, code review, and workflows across your existing coding-agent CLIs.
- [Jackalope vs Superset: Workflow comparison](https://jackalope.dev/compare/superset/): Jackalope vs Superset: compare parallel coding agents, Git worktrees, recurring tasks, remote hosts, accounts, and how changes reach review.
- [Jackalope vs Orca: Workflow comparison](https://jackalope.dev/compare/orca/): Jackalope vs Orca: compare agent workspaces, browser evidence, terminals, remote access, account workflows, and review of parallel changes.
- [Jackalope vs Conductor: Workflow comparison](https://jackalope.dev/compare/conductor/): Jackalope vs Conductor: compare local agent coordination with cloud workspaces, shared sessions, agent subscriptions, and code review workflows.
- [Jackalope vs Emdash: Workflow comparison](https://jackalope.dev/compare/emdash/): Jackalope vs Emdash: compare parallel agents, reusable context, MCP tools, recurring tasks, browser previews, remote development, and review.
- [Jackalope vs Braid: Workflow comparison](https://jackalope.dev/compare/braid/): Jackalope vs Braid: compare Git worktrees, agent sessions, project setup, GitHub review, task dependencies, and integration of parallel changes.
- [Jackalope vs Worktree: Workflow comparison](https://jackalope.dev/compare/worktree/): Jackalope vs Worktree: compare local agent coordination with shared AI threads, team collaboration, cloud execution, MCP connections, and review.
- [Jackalope vs Codius: Workflow comparison](https://jackalope.dev/compare/codius/): Jackalope vs Codius: compare local agent tasks with desktop, web, and mobile access, host ownership, recurring work, providers, and code review.
- [Jackalope vs Hermes Agent: Workflow comparison](https://jackalope.dev/compare/hermes/): Jackalope vs Hermes Agent by Nous Research: compare coding-agent workspaces with persistent agent memory, messaging, automation, and task review.
- [Parallel coding agents with one review workflow | Jackalope](https://jackalope.dev/parallel-coding-agents/): Run Codex, Claude Code, Grok, OpenCode, Kimi Code, Gemini CLI, and Antigravity in parallel worktrees with shared project context and review in Jackalope.
- [Git worktrees for AI coding agents | Jackalope](https://jackalope.dev/git-worktrees-for-ai-agents/): Learn how Jackalope uses isolated Git worktrees to run AI coding agents in parallel and guards review, integration, recovery, and cleanup.
- [Supported AI coding agents | Jackalope](https://jackalope.dev/agents/): Compare Codex, Claude Code, Grok, OpenCode, Kimi Code, Gemini CLI, and Antigravity support for tasks, tools, accounts, and usage in Jackalope.
- [Gemini CLI desktop workspace and GUI | Jackalope](https://jackalope.dev/agents/gemini-cli/): Use Gemini CLI in Jackalope’s desktop workspace for tasks, Git worktrees, saved conversations, and review. Check setup, permissions, and current limits.
- [Run Codex in a desktop project workspace | Jackalope](https://jackalope.dev/agents/codex/): Run Codex tasks in isolated Git worktrees with project context, MCP connections, account profiles, reported usage, and review in Jackalope.
- [Run Claude Code in a desktop project workspace | Jackalope](https://jackalope.dev/agents/claude-code/): Run Claude Code tasks with isolated Git worktrees, project MCP connections, in-app questions, account profiles, usage reporting, and review.
- [Run Grok in isolated Git worktrees | Jackalope](https://jackalope.dev/agents/grok/): Run Grok coding-agent tasks in isolated Git worktrees with project context, on-demand HTTP tool discovery, task history, and review in Jackalope.
- [Run OpenCode tasks in isolated Git worktrees | Jackalope](https://jackalope.dev/agents/opencode/): Run OpenCode tasks in isolated Git worktrees with project context, task history, coordination, and review while retaining OpenCode CLI configuration.
- [Use Kimi Code for project tasks | Jackalope](https://jackalope.dev/agents/kimi-code/): Set up Kimi Code in Jackalope with separate accounts, model choices, task approvals, and session continuation. Check tool, account, and usage coverage.
- [How to run Codex and Claude Code in parallel | Jackalope](https://jackalope.dev/guides/run-codex-and-claude-code-in-parallel/): A practical workflow for running Codex and Claude Code in separate Git worktrees, coordinating dependencies, and reviewing their combined changes.
- [How to review AI-generated code | Jackalope](https://jackalope.dev/guides/review-ai-generated-code/): Review AI-generated code with the original brief, exact patch snapshot, checks, evidence, follow-ups, and combined integration state in view.
- [Project context for AI coding agents | Jackalope](https://jackalope.dev/features/project-context-for-coding-agents/): Give coding-agent tasks selected project instructions, lessons, reusable workflows, history, and MCP tools and inspect what each task receives.
- [Recurring coding-agent tasks & review history | Jackalope](https://jackalope.dev/features/recurring-coding-agent-tasks/): Schedule recurring coding-agent work with its project, account, context, missed-run policy, isolated worktree, result, and review history intact.
- [Coding agent bots with roles and schedules | Jackalope](https://jackalope.dev/blog/bots-that-know-their-job/): Save a coding agent as a bot with standing instructions, an agent, a project and its tools. Ask it questions, give it work, or let it run on a schedule.
- [Run AI coding agents from the terminal | Jackalope](https://jackalope.dev/blog/coding-agents-from-the-terminal/): Start routed coding-agent conversations from any Git repository, answer agent questions in place, and move a conversation between the app and your terminal.
- [From coding-agent chat to a reviewed change | Jackalope](https://jackalope.dev/blog/chat-to-reviewed-code-change/): Queue follow-ups, review one combined result, and keep approval separate from merging in Jackalope’s latest task and chat workflows.
- [Clearer handoffs for parallel coding agents | Jackalope](https://jackalope.dev/blog/parallel-work-clearer-handoffs/): Keeping the whole request in view, passing on checked work, and bringing related changes back for one considered review.
- [Focused agent guidance and useful checks | Jackalope](https://jackalope.dev/blog/focused-guidance-useful-checks/): What belongs in a task brief, what belongs in a check result, and why fewer tokens alone cannot tell us if an agent did better.
- [Smoother task history and code review | Jackalope](https://jackalope.dev/blog/making-room-for-longer-histories/): A look at the work behind incremental saves, more selective interface updates, and large patches that leave room to keep working.
- [AGENTS.md vs CLAUDE.md: a practical guide | Jackalope](https://jackalope.dev/blog/agents-md-claude-md-project-instructions/): Write an AGENTS.md or CLAUDE.md coding agents follow: what to include, sharing one file across Codex, Claude Code and Gemini CLI, and testing it.
- [Git worktree tutorial for coding agents | Jackalope](https://jackalope.dev/blog/git-worktrees-for-ai-coding-agents/): A practical Git worktree workflow for coding agents, with setup commands, separate dev servers, review checks, and careful cleanup.
- [Using Claude Code and Codex together | Jackalope](https://jackalope.dev/blog/claude-code-and-codex-together/): Plan independent tasks, run Claude Code and Codex in separate workspaces, exchange review evidence, and check the combined result.
- [AI code review checklist with examples | Jackalope](https://jackalope.dev/blog/review-ai-generated-code-checklist/): Review AI-generated code against requirements, meaningful tests, and the combined patch. Includes examples and a reusable review brief.
- [MCP for coding agents: tools and permissions | Jackalope](https://jackalope.dev/blog/mcp-for-coding-agents/): Build a focused MCP setup for coding agents. Choose useful tools, understand access, test connections, and keep project context deliberate.
- [How to choose an AI coding-agent workspace | Jackalope](https://jackalope.dev/blog/choose-ai-coding-agent-workspace/): Choose a coding-agent workspace around task setup, parallel work, review, tools, and recovery. Includes a practical evaluation scorecard.
- [Help shape a coding-agent workspace | Jackalope](https://jackalope.dev/blog/building-for-your-workflow/): Tell us which agents and workflows matter to you, and bring other developers into early access.
- [Separate work and personal coding-agent accounts | Jackalope](https://jackalope.dev/blog/work-and-personal-accounts/): Set up separate agent sign-ins, choose accounts for each project, and inspect task usage together in Jackalope.
- [Why we are building a coding-agent workspace | Jackalope](https://jackalope.dev/blog/room-for-the-work/): Why we are building a calmer desktop home for your projects, coding agents, and the decisions that stay yours.
- [Coding-agent workflow: From brief to code review | Jackalope](https://jackalope.dev/blog/from-brief-to-review/): A practical walkthrough of the task workflow: choose the project, describe the outcome, then inspect what comes back.
- [Getting started | Jackalope](https://jackalope.dev/knowledge/getting-started-in-the-beta/): Install Jackalope, connect your account, find your way around and send feedback while the app is in beta.
- [The Work page | Jackalope](https://jackalope.dev/knowledge/workspace-modes/): The sidebar keeps every piece of work in view, a board shows everything in progress, and the sidebar folds away when you want one thing on screen.
- [Bots | Jackalope](https://jackalope.dev/knowledge/bots/): Save an agent with a role, instructions and the tools it may use, then talk to it, let it wake up on its own and work with other bots.
- [Terminal command | Jackalope](https://jackalope.dev/knowledge/terminal-command/): Use the jackalope command to start and rejoin conversations from the current directory, and move a conversation between the app and your terminal.
- [Conversations and follow-ups | Jackalope](https://jackalope.dev/knowledge/chat-and-follow-ups/): Start a conversation, queue corrections, pause work, and review a saved result in Jackalope. Recover after a failed attempt or restart.
- [Review and merge AI changes | Jackalope](https://jackalope.dev/knowledge/review-and-merge/): Inspect changed files and checks, approve a task outcome, and merge into your target branch. Learn why a merge can be blocked and how to recover.
- [Ask Jackalope | Jackalope](https://jackalope.dev/knowledge/ask-jackalope/): Use your configured agent to find answers, personalize Jackalope and prepare work. Connect external agents to the same local tools.
- [Git worktrees | Jackalope](https://jackalope.dev/knowledge/git-worktrees/): How Jackalope uses native Git worktrees to isolate parallel coding agents, prevent dirty checkout collisions, and enforce guarded fast-forward integration.
- [Cloudflare Artifacts | Jackalope](https://jackalope.dev/knowledge/cloudflare-artifacts/): Connect your Cloudflare account, create a project backed by an Artifacts repository or move an existing one, push branches, and share expiring review snapshots.
- [Routing & usage limits | Jackalope](https://jackalope.dev/knowledge/task-routing-and-quotas/): Choose eligible agents using reported capacity, preserve work during quota handoff, and understand pinned tasks and limits.
- [Agents & accounts | Jackalope](https://jackalope.dev/knowledge/multi-account-and-agents/): Set up supported agent CLIs, choose project accounts, and understand credential storage limits.
- [Tools & browser automation | Jackalope](https://jackalope.dev/knowledge/mcp-and-browser-automation/): Choose tools for a task, check agent compatibility, and inspect browser results with useful evidence.
- [Windows desktop control | Jackalope](https://jackalope.dev/knowledge/windows-desktop-control/): Select a window for a task, understand its control status, and pause or cancel when you need your computer.
- [Troubleshooting & diagnostics | Jackalope](https://jackalope.dev/knowledge/troubleshooting-and-diagnostics/): Narrow a failure to setup, routing, execution, or recovery, then collect the right evidence for support.
- [Fix CLI discovery | Jackalope](https://jackalope.dev/knowledge/fixing-cli-path-on-windows/): Fix Windows CLI detection for Codex, Claude Code, Grok, OpenCode, Kimi Code, and Antigravity with PATH checks and restart steps.
- [Resolve Git locks | Jackalope](https://jackalope.dev/knowledge/resolving-git-worktree-locks/): How to safely diagnose, remove, and prevent stale Git index locks and orphaned worktrees left by abnormal system termination.
- [Connect MCP tools | Jackalope](https://jackalope.dev/knowledge/connecting-custom-mcp-servers/): Add local commands or remote HTTP endpoints, choose the project and agent, and verify a real tool request.
- [Recurring tasks | Jackalope](https://jackalope.dev/knowledge/recurring-schedules-and-automation/): Create a repeatable check, choose a timezone and missed-run policy, and inspect each scheduled result.
- [Your first task | Jackalope](https://jackalope.dev/knowledge/task-composer-and-effort-levels/): Describe outcomes, match project guidance, choose effort, and review changes with their saved evidence.
- [Appearance & themes | Jackalope](https://jackalope.dev/knowledge/theme-editor-and-atmosphere/): Preview colors, appearance, and atmosphere, keep a comfortable theme, and understand companion activity.
## Optional
- [Full text](https://jackalope.dev/llms-full.txt)
- [RSS feed](https://jackalope.dev/feed.xml)
