Pixel Agents is an open source tool that turns Claude Code agents running in your terminal into characters working in a pixel art office. Each agent is assigned one character, and what it is doing is shown through animation at a glance: typing at a desk while writing code, reading while searching, a speech bubble overhead while waiting for approval. It is built by Pablo De Lucca. It first shipped as a VS Code extension, and a standalone version (CLI plus browser app) distributed via npm was added later. The license is MIT.
When you run several agents in parallel, it becomes hard to tell which one has stalled and which one is waiting for your reply. Pixel Agents replaces that status check ─ normally a matter of reading logs ─ with simply looking at an office. It is less a serious progress management tool than a “visualization layer” that makes agent activity enjoyable to follow.
Key Features
- Real-time display with one character per agent: Each agent sits at a desk and acts out what it is doing ─ typing while editing files, reading while searching. Six characters are available to choose from
- Approval waits shown as speech bubbles: When an agent is waiting for input, a speech bubble appears. Optional sound notifications can also be enabled, so you notice a waiting agent even while working on something else
- Visualization of sub-agents and Agent Teams: Sub-agents and Agent Teams are drawn as separate characters, so you can see the whole picture of parallel execution including parent-child relationships
- Office layout editor: You can place floors, walls, furniture, and carpets yourself, and the office you build can be saved and shared. Named “Areas” can be mapped to workspace folders for separation
- Claude Code Hooks integration: Session management is hooks-first, falling back to polling in environments where hooks are unavailable. External sessions started outside VS Code are also picked up
- Both VS Code extension and standalone versions: In addition to installing from the VS Code Marketplace or Open VSX, it can be launched as a browser app with
npx pixel-agents
Pricing
| Plan | Price | Key features |
|---|---|---|
| Open source (MIT license) | $0 | All features available for free. No paid plans or subscriptions are offered |
Pricing is as of August 2026. Please check the official repository for the latest information.
Pros & Cons
✅ Pros
- Grasp the state of agents running in parallel at a glance, without chasing logs
- Less likely to overlook an agent waiting for approval, reducing wasted waiting time
- Completely free under the MIT license. No account registration or payment required to try it
- Available as both a VS Code extension and an npm standalone version, so it works regardless of your editor
- The office layout and characters can be customized to your taste
⚠️ Cons
- Requires the Claude Code CLI to be installed; it does not work on its own
- Claude Code is currently the only supported agent (support for Codex, Gemini, Cursor and others is still on the roadmap)
- Focused on visualization; it has no orchestration features such as task assignment, branch management, or diff review
- Requires VS Code 1.105.0 or later, or Node.js 20 or later for the standalone version
- As an individually developed project, update frequency and support are not on par with commercial tools
Comparison with Similar Services
| Item | Pixel Agents | Claude Squad | Vibe Kanban | Conductor |
|---|---|---|---|---|
| Main use | Visualizing agent status | Terminal session management | Kanban-style task management | Parallel execution and diff review |
| Form factor | VS Code extension / browser app | CLI (tmux based) | Web app | macOS desktop app |
| Supported platforms | Windows / Linux / macOS | Windows / Linux / macOS | Cross-platform | macOS only |
| Task management | None (visualization only) | Managed per session | Managed on a kanban board | Managed per worktree |
| License | Open source (MIT) | Open source | Open source | Commercial |
Who Is It For
- People who run Claude Code sub-agents or Agent Teams in parallel every day and find status tracking tedious
- People who tend to lose time because they overlook agents waiting for approval
- People who would rather grasp the whole picture visually than follow terminal logs
- People who want to add a bit of playfulness to their workspace, or show “agents at work” in a team demo or stream
- People who want to start with a free, open source tool
Summary
Pixel Agents turns multi-agent execution in Claude Code from information you read into information you see. It is not a full orchestration platform that handles task assignment and review; it is a visualization layer for understanding status. On that single point, though, the setup cost is close to zero and the benefit is easy to see. If you already use the Claude Code CLI, it is worth a few minutes to try it from the VS Code Marketplace or with npx pixel-agents.