llms.txt
# Parallel Code
> Parallel Code is a free, open-source (MIT) desktop app for running several AI coding agents — Claude Code, Codex CLI, Gemini CLI, Copilot CLI, and Antigravity CLI — in parallel, each isolated in its own git worktree. Builds for macOS and Linux.
Key features: parallel agents with automatic git-worktree isolation, Arena mode (race agents head-to-head on one task and keep the winning diff), a diff-first review surface, a built-in chat view for Claude Code and Codex, guided tours of plans, files and changes, per-task mind maps and reasoning graphs, agent-initiated child tasks with review, bring-your-own editor and agent, and mobile progress monitoring via QR code over Wi-Fi or Tailscale. Runs locally with your own API keys.
## Comparisons
- [Parallel Code vs Claude Squad](https://parallelcode.app/compare/parallel-code-vs-claude-squad/): Parallel Code vs Claude Squad compared — a native desktop GUI vs a tmux terminal app, licensing, agents and Arena mode, plus when each fits best.
- [Parallel Code vs Superset](https://parallelcode.app/compare/parallel-code-vs-superset/): Parallel Code vs Superset compared — genuine open source (MIT) vs Elastic License, free phone monitoring, plus when each fits best.
- [Parallel Code vs Vibe Kanban](https://parallelcode.app/compare/parallel-code-vs-vibe-kanban/): Looking for a Vibe Kanban alternative? Compare local workspaces, agent support, Arena, and phone monitoring after its project-board retirement.
- [Parallel Code vs Conductor](https://parallelcode.app/compare/parallel-code-vs-conductor/): Looking for a Conductor alternative on Linux or macOS? Compare Parallel Code's agents, worktrees, open-source license, pricing, and review workflow.
## Guides
- [Prevent AI Coding Agents Pushing to Main on GitHub](https://parallelcode.app/blog/protect-main-from-agent-pushes/): Set GitHub branch rules that require pull requests, passing CI checks, and review before an AI coding agent’s changes can reach main.
- [Git Sparse Checkout Per Worktree for AI Agents](https://parallelcode.app/blog/sparse-checkout-per-agent-worktree/): Set distinct sparse checkout directories for each Git worktree, verify each agent's view, and handle shared files, dependencies, and config safely.
- [Copilot CLI: New Conversation in a Worktree](https://parallelcode.app/blog/copilot-cli-new-worktree-conversation/): Learn how to start a second Copilot CLI conversation in a new Git worktree, keep the first session running, and review and merge the results.
- [Verify an Agent Is in the Correct Git Worktree](https://parallelcode.app/blog/verify-agent-worktree-and-branch/): Learn the Git commands that confirm an AI agent's worktree path, checked-out branch, repository identity, and pending changes before edits or commits.
- [Is Git Stash Shared Between Worktrees?](https://parallelcode.app/blog/git-stash-shared-agent-worktrees/): Learn why Git stash is shared across worktrees, how one agent can apply another's changes, and how to keep unfinished work separate.
- [AI Agent Worktree index.lock Permission Denied](https://parallelcode.app/blog/agent-worktree-git-index-lock-permissions/): Learn why an agent can edit a Git worktree but cannot stage or commit, how to locate the blocked index.lock, and which fixes preserve sandboxing.
- [Git Worktree Submodules Not Initialized: How to Fix It](https://parallelcode.app/blog/git-submodules-ai-agent-worktrees/): Learn why submodule directories are empty in a new agent worktree, how to initialize them, and what to check when files are still missing.
- [Claude Code Worktree From Current Branch](https://parallelcode.app/blog/claude-code-worktree-base-branch/): Learn why Claude Code worktrees can start from main, how to inherit your feature branch’s commits, verify the base, and fix an existing worktree.
- [Git Hooks Not Running in a Worktree? Fix the Path](https://parallelcode.app/blog/git-hooks-ai-agent-worktrees/): Find why Git hooks stop running in agent worktrees, inspect the effective hooks path, and configure shared or per-worktree hooks safely.
- [Parallel Coding Agents and Duplicate MCP Servers](https://parallelcode.app/blog/limit-mcp-servers-parallel-agents/): Learn why parallel agent sessions launch duplicate MCP server processes and how to disable unneeded servers in Claude Code, Codex CLI, and Gemini CLI.
- [How to Detect When Codex Exec or Claude Code Finishes](https://parallelcode.app/blog/detect-headless-coding-agent-completion/): Learn to distinguish completed, failed, and stalled Codex exec or Claude Code headless runs using process status, JSON events, time limits, and review checks.
- [Cherry-Pick a Commit Between Git Worktrees](https://parallelcode.app/blog/cherry-pick-between-agent-worktrees/): Learn how to cherry-pick one completed fix between Git worktrees, resolve conflicts, check dependencies, and keep both agents' branches separate.
- [How to Update a Git Worktree Branch From Main](https://parallelcode.app/blog/update-agent-worktree-from-main/): Learn to update a Git worktree branch from main during an active agent task, protect unfinished edits, resolve conflicts, and review the result.
- [Git Worktree Branch Already Checked Out: Safe Fixes](https://parallelcode.app/blog/agent-worktree-branch-already-checked-out/): Fix the Git worktree branch already checked out error: find the owning worktree, inspect agent edits, and safely reuse or release the branch.
- [Why AI Agent Worktrees Miss Uncommitted Changes](https://parallelcode.app/blog/agent-worktree-uncommitted-changes/): Learn why AI agent worktrees miss edits in your main checkout, then use a commit, stash, or patch to give an agent the changes it needs.
- [Gemini CLI --worktree for Parallel Sessions](https://parallelcode.app/blog/gemini-cli-parallel-worktrees/): Enable Gemini CLI worktrees, launch concurrent sessions in separate directories, and review, merge, resume, and clean up their changes.
- [Resume a Coding Agent Session in the Right Git Worktree](https://parallelcode.app/blog/resume-agent-session-worktree/): Learn how to find the right Git worktree, check its branch and pending changes, then resume a Claude Code, Codex, or Gemini CLI session safely.
- [Parallel AI Coding Agents: Handle 429 Rate Limits](https://parallelcode.app/blog/parallel-ai-agent-rate-limits/): Handle 429 rate limits in parallel AI coding agents: identify shared quotas, stagger sessions, and retry temporary errors without wasting requests.
- [Clean Up Stale AI Agent Git Worktrees Safely](https://parallelcode.app/blog/clean-up-ai-agent-worktrees/): Learn to clean up stale AI agent Git worktrees by auditing files and commits, preserving unfinished work, and removing worktrees and branches safely.
- [Parallel AI Agents: Stop Sharing a Test Database](https://parallelcode.app/blog/isolate-test-databases-parallel-ai-agents/): Learn how to give each parallel AI agent its own test database, run migrations safely, isolate Compose stacks, and verify connections before tests.
- [AGENTS.md and CLAUDE.md: One Context File for Every Agent](https://parallelcode.app/blog/agents-md-vs-claude-md/): Claude Code, Codex, Gemini CLI, and Copilot CLI read different instruction files. How to keep one source of truth without duplicating it four ways.
- [Is It Safe to Run AI Coding Agents With Permissions Off?](https://parallelcode.app/blog/ai-agent-permissions-and-sandboxing/): Skipping approval prompts is not the same as sandboxing. What each agent's permission and sandbox settings actually enforce, and what parallel runs change.
- [How to Merge Work From Parallel AI Agents](https://parallelcode.app/blog/merging-parallel-ai-agent-branches/): Parallel agents finish on separate branches. A practical integration workflow: test the merges before you run them, pick an order, and keep main releasable.
- [How to Run Multiple Claude Code Agents in Parallel](https://parallelcode.app/blog/multiple-claude-code-agents-parallel/): Run multiple Claude Code agents with worktrees, agent view, subagents, cloud sessions, or Parallel Code—and know when to choose each approach.
- [How to Use Claude Code and Codex Together: 4 Practical Workflows](https://parallelcode.app/blog/claude-code-and-codex-together/): Use Claude Code and Codex together for parallel tasks, cross-review, handoffs, and races—with isolated working copies when both agents write.
- [How to Run Multiple Codex Agents in Parallel](https://parallelcode.app/blog/multiple-codex-agents-parallel/): Run multiple Codex agents using the desktop app, CLI worktrees, subagents, cloud tasks, or Parallel Code—and know when to choose each approach.
- [How to Write Better Prompts for AI Coding Agents](https://parallelcode.app/blog/ai-coding-agent-prompts/): A practical AI coding agent prompt template for clearer scope, focused diffs, stronger verification, and fewer avoidable retries.
- [AI Coding Agent Best Practices: A Practical Workflow](https://parallelcode.app/blog/ai-coding-agent-best-practices/): Use AI coding agents without losing control. These 10 best practices cover task scope, worktree isolation, tests, review, security, cost, and merging.
- [What Does Running Multiple AI Agents in Parallel Actually Cost?](https://parallelcode.app/blog/cost-of-running-multiple-ai-agents/): Learn how subscriptions, metered APIs, repeated context, and redundant races affect the real cost of running multiple AI coding agents.
- [Running AI Coding Agents on Linux: Your Real Options](https://parallelcode.app/blog/ai-coding-agents-on-linux/): Most AI agent orchestrators are macOS-only. Here's what actually runs on Linux — native GUIs, terminal tools, and the CLIs — and how to pick one.
- [Racing AI Agents: Run Claude, Codex, and Gemini on the Same Task](https://parallelcode.app/blog/racing-ai-agents-same-task/): Stop guessing which agent is best. Run several on the same task in parallel, then keep the winning diff. How to race AI coding agents and judge the results.
- [How to Split a Feature Into Parallel AI Agent Tasks](https://parallelcode.app/blog/splitting-work-for-parallel-ai-agents/): Parallel agents only save time when tasks don't overlap. A practical guide to splitting a feature into independent tasks you can run at the same time.
- [Multi-Agent Coding Tools in 2026: An Honest Comparison From Someone Who Built One](https://parallelcode.app/blog/multi-agent-coding-tools-2026/): How to choose between Parallel Code, Nimbalyst, Conductor, Claude Squad, Vibe Kanban, Augment Intent, Gas Town, and Antfarm.
- [How to Use Multiple AI Coding Agents on One Repo](https://parallelcode.app/blog/multiple-ai-agents-one-repo/): A practical guide to running Claude Code, Codex CLI, and Gemini CLI on one codebase without working-copy collisions or merge chaos.
- [Claude Code vs Codex vs Gemini CLI Compared](https://parallelcode.app/blog/claude-code-vs-codex-vs-gemini/): A practitioner's comparison of the three major AI coding agents — code quality, cost, and speed — with guidance on when to use each.
- [AI Agent Isolation: Separate Working Copies for Concurrent Tasks](https://parallelcode.app/blog/parallel-ai-agents/): Why every parallel AI coding agent needs its own git worktree, branch, and working directory before you scale beyond one session.
- [Git Worktrees: Why AI Agents Need Them](https://parallelcode.app/blog/git-worktrees-explained/): Git worktrees let you check out multiple branches simultaneously from one repo. Here's how they work and why they matter for parallel development.
- [How to Review AI-Generated Code Efficiently](https://parallelcode.app/blog/reviewing-ai-generated-code/): AI agents write code fast, but reviewing it is the real bottleneck. A practical checklist and workflow for staying in control.
## More
- [Parallel Code (home)](https://parallelcode.app/): Free, open-source desktop app for running AI coding agents in parallel.
- [About the creator](https://parallelcode.app/about/): Authorship, project background, and editorial approach.
- [Tool comparisons](https://parallelcode.app/compare/): How Parallel Code compares to other multi-agent coding tools.
- [Blog](https://parallelcode.app/blog/): Guides on parallel AI agents, git worktrees, and multi-agent workflows.
Sitemap: https://parallelcode.app/sitemap-index.xml
