omacom/omarchy

Selection

Select when

• You want a complete Linux environment rather than a Claude Code add-on.
• You want Claude provisioned alongside several alternative coding-agent CLIs.
• You want Claude limits and local usage visible in the desktop bar.
• You want one UI to compare multiple enabled coding-agent subscriptions.
• You accept an opinionated Hyprland and Quickshell desktop architecture.
• You want a manifest-based plugin system for desktop widgets and services.
• You need optional cross-device aggregation of local agent-usage snapshots.
• You are prepared to review unsandboxed third-party QML plugins before enabling them.
• You want desktop controls and agent monitoring in one distribution.
• You can dedicate a machine or installation to this workstation design.

Boundaries

Avoid when

• You only need a portable Claude Code skill, hook, plugin, or MCP server.
• You need a solution that installs inside Claude Code without replacing the workstation environment.
• You require third-party desktop plugins to run in a security sandbox.
• You cannot permit collectors to inspect local agent transcripts or credential-associated endpoints.
• You need independently verified billing or token accounting.
• You want a distribution-neutral package for macOS, Windows, or an existing Linux desktop.
• You require a stable interface with a supplied version-compatibility matrix.
• You cannot tolerate opinionated system, shell, desktop, and service configuration.
• You need low-effort setup or easy removal from an existing workstation.
• You require all agent features to work fully offline.

Strengths

Capabilities

Risk profile

Risks and limitations

• Uncertain: only selected first-party files were supplied, so the installer, collectors, sandbox boundaries, update process, and broader manual were not fully inspected.
• Uncertain: supplied update, push, and embedded documentation dates extend into 2026, so current release state and recency cannot be independently reconciled.
• Uncertain: no first-party compatibility matrix establishes supported Claude CLI or Claude Code versions.
• Uncertain: the provided files show a Claude installation entry but do not explain exactly when the stub resolves or which package and version it installs.
• Third-party Omarchy plugins execute unsandboxed inside the long-running desktop shell and therefore require source review and trust.
• The menu can evaluate bash expressions and execute configured action strings, increasing the impact of untrusted configuration.
• Claude account limits require a signed-in CLI, and failures fall back to local statistics only.
• Usage completeness depends on retained local transcripts, fallback files, provider records, and collector-specific lookback windows.
• Codex native sessions and Fireworks billing data are limited to a stated 30-day window for some totals.
• Cross-device aggregation relies on an external synchronized directory and does not itself provide or secure the transport.
• Adoption entails a complete, opinionated Linux environment rather than a narrowly scoped Claude enhancement.
• No independent security audit, privacy assessment, usage-accuracy benchmark, or reproducible compatibility test was supplied.

Documented third-party plugins are git repositories loaded as unsandboxed QML code inside omarchy-shell. The plugin workflow warns before cloning, installs additions disabled, validates manifests, shows update diffs, and states that installation itself runs no plugin code, hooks, or sudo. However, enabled code shares a long-running desktop process, and the menu evaluates configured bash expressions and executes selected action strings. Supplied paths also indicate firewall, lock-screen PAM, Polkit, snapshots, and authentication work, but their implementations were not inspected. Inference: the security posture depends heavily on trusting the distribution, reviewing plugins and configuration, and protecting local agent credentials and records.