Key takeaways
- Task queue system for dispatching work to multiple AI coding agents with priority reordering
- Unified approval inbox for reviewing agent permission requests across all running agents
- Open source (github.com/scarce/axel, ~204 stars as of June 2026) and free, built by Txtx, Inc.; native macOS, iOS, and visionOS apps
FAQ
What is Axel?
Axel is a free, open-source native Mac task manager for AI coding agents from Txtx, Inc.—queue tasks, dispatch them to the right agent, and approve or deny permission requests from one inbox.
What agents does Axel support?
Axel works with Claude, Codex, OpenCode, and Antigravity out of the box.
How is Axel different from other agent apps?
Axel focuses on task queuing and approval governance rather than being another IDE; it's 'Todoist for AI coding agents.'
Executive Summary
Axel positions itself as "Todoist for AI coding agents"—a task queue and approval inbox that lets developers dispatch work to multiple AI agents and manage permission requests from a single interface. Unlike IDE-style agent apps, Axel focuses specifically on orchestration, queuing, and governance rather than code editing.
Status (June 2026): Alive. The website is live with active downloads, the app is free and open source (the repository moved from txtx/axel-app to scarce/axel), and the Product Hunt launch on February 8, 2026 drew 284 upvotes and a #3 day rank. Repo activity has slowed, however—the last push was April 28, 2026.
| Attribute | Value |
|---|---|
| Company | Txtx, Inc. (maker: Ludo Galabru) |
| Founded | 2026 (launched February 2026) |
| Funding | Not publicly disclosed |
| Employees | Unknown |
| Headquarters | Unknown |
Product Overview
Axel addresses the operational friction of running multiple AI coding agents. Instead of juggling separate terminal windows or browser tabs for different agents, developers use Axel to queue tasks, assign them to the appropriate agent (Claude, Codex, OpenCode, or Antigravity), and handle all permission requests from a unified inbox.
The app uses a declarative approach: an AXEL.md file defines your workspace layout, panes, skills, and grid positions. Tasks can be reordered while running, and sessions persist via tmux or iTerm2.
Key Capabilities
| Capability | Description |
|---|---|
| Task Queue | Add tasks, assign to agents, reorder priorities on the fly |
| Unified Inbox | Approve/deny file edits, commands, API calls from all agents |
| Portable Skills | Skills in ~/.config/axel/skills symlinked to each agent |
| Worktree Automation | axel -w feat/auth spawns worktree + tmux session |
| Auto-Approve Rules | Skip inbox for read-only ops or small edits under N tokens |
| Token & Cost Tracking | Per-task token and cost reporting |
| macOS Notifications | Get pinged when an agent is blocked waiting for approval |
Product Surfaces
| Surface | Description | Availability |
|---|---|---|
| macOS App | Native SwiftUI with menu bar | GA |
| iOS App | Mobile task management | GA |
| visionOS App | Spatial computing support | GA |
Technical Architecture
Axel manages agent sessions through tmux or iTerm2, allowing sessions to persist after closing the terminal. Workspaces are defined declaratively in AXEL.md with YAML frontmatter specifying panes, skills, and layout. Git worktrees provide isolation for parallel agent work.
Key Technical Details
| Aspect | Detail |
|---|---|
| Deployment | Local Mac/iOS/visionOS app |
| Session Management | tmux or iTerm2 |
| Workspace Config | AXEL.md with YAML frontmatter |
| Isolation | Git worktrees |
| Open Source | Yes — scarce/axel; ~204 stars, 15 forks as of June 2026; no license declared in repo metadata |
Strengths
- Focused scope — Does one thing well: task queuing and approval governance; not trying to be an IDE
- Multi-agent dispatch — Same queue, different agents; pick the right tool for each task
- Keyboard-driven — New pane, dispatch, reorder, kill—all via shortcuts; Spotlight integration
- Approval governance — Centralized inbox with full context (file path, diff preview, command args) before approving
- Cross-platform Apple — Native apps for macOS, iOS, and visionOS
- Session persistence — tmux-backed sessions survive terminal closure
Cautions
- Apple-only — No Windows or Linux support; reviewers flag the lack of cross-platform support and team collaboration as gaps
- Slowing development — The GitHub repo's last push was April 28, 2026, roughly six weeks before this update; only one tagged release (cli-v0.8.0, March 2026)
- Thin community feedback — 284 Product Hunt upvotes but zero reviews on the page; no Hacker News or Reddit discussion found as of June 2026
- Configuration overhead — AXEL.md and skills setup may have learning curve
- Worktree conflicts — Multiple agents touching overlapping files can create merge conflicts
- No declared license — The repo is public but GitHub metadata shows no license, leaving reuse terms ambiguous
What Developers Say
Community discussion is thin: Axel has no Hacker News threads (zero results on hn.algolia.com for axel.build) and no substantive Reddit discussion as of June 11, 2026. The Product Hunt launch page carries the only public feedback.
- Praise (Product Hunt): One commenter called the worktree design smart, saying spawning agents is "as cheap as opening a tab."
- Praise (Product Hunt): Another appreciated that parallel dispatch solved the friction of "I can't grab the task to start next or work on two different projects."
- Maker framing (Product Hunt): Creator Ludo Galabru describes it as "Not an agent that happens to have a task list. A task list that happens to command agents."
- Criticism (FunBlocks review): The reviewer flags missing monitoring dashboards, team collaboration features, and the macOS-only limitation.
Pricing & Licensing
| Tier | Price | Includes |
|---|---|---|
| Free | $0 | Full app functionality (confirmed on live site, June 2026) |
Licensing model: Free, open-source app; BYOK for agents. No paid tiers or subscriptions as of June 2026.
Hidden costs: Agent subscriptions (Claude, Codex, etc.) paid separately
Competitive Positioning
Direct Competitors
| Competitor | Differentiation |
|---|---|
| Agentastic | Axel focuses on task queuing/approval; Agentastic is a full terminal IDE |
| Auto-Claude | Axel is human-driven dispatch; Auto-Claude is autonomous execution |
| Commander AI | Axel is about orchestration; Commander is about prompt→diff→commit |
When to Choose Axel Over Alternatives
- Choose Axel when: You need task queuing and centralized approval governance across agents
- Choose Agentastic when: You want full terminal access and worktree-based development
- Choose Auto-Claude when: You want fully autonomous agent execution
Ideal Customer Profile
Best fit:
- Developers managing multiple AI agents who want structured task dispatch
- Teams needing approval governance and audit trails for agent actions
- Apple ecosystem users (Mac, iPhone, Vision Pro)
- Power users who prefer keyboard-driven interfaces
Poor fit:
- Developers wanting a full IDE or terminal experience
- Windows or Linux users
- Teams preferring autonomous agent execution without approval gates
Viability Assessment
| Factor | Assessment |
|---|---|
| Financial Health | Backed by Txtx, Inc.; funding not publicly disclosed |
| Market Position | Niche (orchestration-focused) |
| Innovation Pace | Slowing — last repo push April 28, 2026; one tagged release since launch |
| Community/Ecosystem | Small — ~204 GitHub stars, 284 PH upvotes, no HN/Reddit traction as of June 2026 |
| Long-term Outlook | Differentiated but unproven; watch commit cadence |
Axel's focus on orchestration rather than IDE features carves out a unique niche, and being free and open source lowers adoption risk. But four months after launch, traction is modest (~204 stars as of June 2026) and the development pace has visibly cooled since March—typical fast-churn-category signals worth monitoring.
Bottom Line
Axel takes a unique approach in the Mac coding agent app space by focusing specifically on task queuing and approval governance rather than being another IDE. This "Todoist for AI agents" positioning could resonate with developers who already have their preferred editor but want better agent orchestration. As of June 2026 the product is alive, free, and open source under Txtx, Inc., but community traction is thin and repo activity has slowed since March.
Recommended for: Developers who want structured task dispatch and centralized approval management across multiple AI agents—the free, open-source model makes it low-risk to trial
Not recommended for: Users wanting a full IDE experience, Windows/Linux developers, those preferring autonomous agent workflows, or teams needing an actively maintained dependency
Outlook: Differentiated positioning with cross-platform Apple support, but the slowing commit cadence and absent community discussion make long-term maintenance the open question; success depends on whether orchestration becomes a distinct product category
Research by Ry Walker Research • methodology