<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.> pixtuoid addresses the challenge of managing multiple AI coding agents by providing a unique, glanceable terminal-based visualization. It transforms each agent session into a pixel-art character at a desk within a virtual office, allowing users to monitor progress, identify bottlenecks, and understand agent status without constant context switching. This offers a highly intuitive and low-distraction multi-agent dashboard for developers and researchers.

How It Works

The system captures agent activity through two primary mechanisms: a non-blocking Unix-socket hook shim and JSONL transcript watching. These event streams are consolidated and processed by a reducer function that maintains the office's state. This state is then rendered as detailed half-block pixel art within the terminal. The core logic is built using three Rust crates, ensuring minimal dependencies and high performance without relying on external terminal libraries. This architecture guarantees that the visualization process never impedes the agent's execution.

Quick Start & Requirements

Installation is streamlined via package managers: use brew install IvanWng97/pixtuoid/pixtuoid on macOS or npm install -g pixtuoid on any OS. After installation, run pixtuoid install-hooks to set up agent integration. Supported coding agents (e.g., Claude Code, Codex CLI) can then be launched in a separate terminal, with their corresponding characters appearing in the office visualization within seconds. Alternative installation methods, including Cargo, prebuilt binaries, and Debian .deb packages, are detailed in the install guide. Configuration settings, such as themes and desk capacity, are managed via ~/.config/pixtuoid/config.toml, with CLI flags offering runtime overrides.

Highlighted Details

• Multi-Agent Office: Each agent occupies a desk, with additional agents automatically populating new floors.
• Dynamic Visuals: Features animated characters performing actions like typing or sleeping, alongside environmental effects like rain, snow, and sunset lighting.
• Agent Status Indicators: Per-tool monitor glows (e.g., blue for Edit, orange for Bash) and hover-over tooltips provide session duration, tool call counts, and active time percentages.
• Interactive Elements: Includes office pets, a coffee run simulation for idle agents, and ambient "pantry chitchat" between idle characters.
• Robust Integration: A hook-safe design ensures the visualizer's shim always exits successfully (0), preventing it from blocking agent operations.

Maintenance & Community

Information regarding specific maintainers, community channels (like Discord or Slack), or a public roadmap is not detailed in the provided documentation. Contribution guidelines for adding new features or agent adapters are available in a dedicated file.

Licensing & Compatibility

The project is released under the MIT license, permitting broad use and modification. While generally compatible across operating systems, certain AI coding agent integrations are noted as experimental or having limited testing on Windows.

Limitations & Caveats

Support for some agent CLIs on Windows is explicitly marked as experimental, indicating potential instability or incomplete functionality in those specific environments. Further details on known bugs or platform-specific limitations are not elaborated upon.