mrkeyoor.com_
Tue 01 Sept 17:42 UTC
Dev Toolsevaluationupdated 26 Aug 2026

asyar review

Asyar is an open-source desktop launcher for macOS, Windows, and Linux, built with Rust, Tauri, and Svelte. It combines app and file search with clipboard history, scripts, extensions, window controls, AI agents, and MCP servers, so keyboard-heavy users can replace several small utilities with one launcher.

+131stars / 7d
Verdict

Our Asyar build stopped after 9 seconds because Bun was missing, despite the 21-second dependency install succeeding, so source contributors should budget for an extra runtime prerequisite. The released app is worth trying if you want one local launcher to cover search, scripts, AI, and MCP across three desktop operating systems. Skip it if Wayland text expansion is essential or if a smaller permission surface matters more than breadth.

We ran it

Lab card: what happened when we ran asyarScreenshot of asyar (asyar.org)
Install✓ · 21s719 packages · 672 MB
Build✗ · 9s
Testsn/ano test script
Repo1871 files~292,811 lines of source · 42.2 MB · 4 CI workflows

Answers from our run

Does asyar build from source?

Dependencies installed in 21 seconds (719 packages), and the build failed. We cloned commit a80dd42 into a clean Debian container with 3 CPUs and no project-specific setup.

Does asyar have tests you can run?

Not through a standard command: the project exposes no test script or target that our harness could run.

Who should not use asyar?

Wayland users who depend on global snippet expansion: the README says input-heavy features such as Snippets do not work there.

What are the alternatives to asyar?

Ulauncher, Flow Launcher, Raycast. Our Asyar build stopped after 9 seconds because Bun was missing, despite the 21-second dependency install succeeding, so source contributors should budget for an extra runtime prerequisite.

Setup2/5Install passed, but the build stopped when Bun was unavailable
Docs4/5Broad user and contributor guidance, with platform limits stated
Community3/5523 stars and five open issues and PRs with same-day activity
Maturity3/5Wide feature set, but source build and Linux reports need attention

Who it’s for

Developers who want a local desktop launcher with scripts, extensions, and file search.
Linux users looking for a Raycast-style launcher that also supports Windows and macOS.
AI users who want to bring their own provider or Ollama and connect MCP servers.
Extension authors comfortable with a TypeScript SDK and a desktop app built on Tauri.

Who it’s NOT for

Wayland users who depend on global snippet expansion: the README says input-heavy features such as Snippets do not work there.
Teams that require a clean source build from the documented setup alone: our build stopped because Bun was absent.
Users who expect every OpenAI-compatible endpoint to accept the same request fields: issue 684 reports a provider rejecting Asyar's temperature parameter.
Linux users unwilling to troubleshoot WebKit and graphics integration: issue 435 reports an AppImage startup crash on Kubuntu 26.04.
Buyers who need a narrowly scoped launcher: Asyar also manages clipboard data, AI credentials, shell tools, file access, extensions, and optional sync.

Setup reality

Our pnpm install succeeded in 21 seconds, adding 719 packages and using 672 MB. The build failed after 9 seconds. It completed the SDK, then stopped because the extension-builder step could not find Bun on PATH or a local Bun binary. No test script or target was available, so we skipped tests.

End users can install a release or use the shell installer without an account. AI features need credentials for OpenAI, Anthropic, Google, OpenRouter, another compatible endpoint, or a local Ollama service. The AI extension builder specifically uses an Anthropic key.

Source contributors use node setup.mjs, while some MCP and extension-building paths download Bun, uv, or Claude after consent. Wayland needs a desktop-level shortcut, and the README says global input features such as Snippets do not work there.

It replaces a launcher, clipboard manager, and AI shortcut layer

Asyar is a desktop command bar built around Rust, Tauri, and Svelte 5. Search covers applications and files, while built-in commands handle calculations, clipboard history, snippets, scripts, window layouts, and URLs. Extensions add more commands through a TypeScript SDK. The same interface can host AI conversations, silent text transforms, and MCP tools. That breadth is the reason to consider it, and the main reason some people should avoid it.

The app targets macOS, Windows, and Linux rather than treating Linux as an afterthought. Its README lists support for application indexing, file search, clipboard history, backup and restore, and 17 window layouts across the three platforms. Linux has a sharp boundary: the support matrix covers X11, and the project says global input-heavy features such as snippet expansion do not work on Wayland. Wayland users must bind a desktop shortcut to the asyar command just to summon the launcher.

AI can use local models, hosted providers, and MCP tools

Agents can connect to OpenAI, Anthropic, Google, OpenRouter, Ollama, or an OpenAI-compatible endpoint. Asyar supplies 8 built-in tool types covering calculator work, clipboard access, files, shell execution, web fetching, and launcher search. MCP configuration can be imported from Claude Desktop, Cursor, Cline, Continue, and Zed. Write and execution tools trigger permission prompts on first use.

Silent agents are the distinctive workflow. A hotkey can send selected text, clipboard contents, or an argument to a model, then replace the selection, paste at the cursor, copy the answer, or show a small status message. That is useful for repeated edits, but it grants the launcher meaningful access to desktop input and output. The privacy controls therefore matter as much as the model list.

Asyar says it stores data locally, disables telemetry by default, filters clipboard items marked private by the operating system, and deny-lists common password managers. Known token patterns are redacted before storage and before AI provider requests. Sensitive local fields use AES-256-GCM with a master key in the operating system keychain. On Linux systems without Secret Service, the documented fallback is a mode-0600 file plus an in-app warning.

What happened when we ran it

Our sandbox installed 719 pnpm packages in 21 seconds and used 672 MB on disk. The checkout at commit a80dd42 contained 1,871 files, about 292,811 lines of source, and 42.2 MB before dependencies. The monorepo had 4 CI workflow files, no Dockerfile, and no tests directory. No test script or target was available, so we did not invent a substitute test command.

The build failed with exit code 1 after 9 seconds. TypeScript compiled the SDK and its CLI successfully. The next extension-builder JavaScript step stopped with a direct message: no bun executable was on PATH, and no local platform-specific Bun binary existed under src-tauri/binaries. The log advised installing Bun or placing the binary manually. That is the complete finding; the log does not establish why the expected runtime was absent.

This matters because the contributor instructions present node setup.mjs as the full development setup. Elsewhere, the README explains that Asyar no longer bundles Bun, uv, or Claude at build time and downloads them on demand for certain runtime features after user consent. Our 9-second failure shows that the source build path still needs Bun available at the point the extension builder is compiled.

The extension system is capable, with a large trust boundary

Extension authors get a CLI for scaffolding, development, building, and publishing. Commands can declare schedules between 60 and 86,400 seconds, publish live subtitle updates into search results, expose actions at the extension or command level, and accept deep links. An in-app AI builder can scaffold and verify an extension using the user's Anthropic key. The README says extensions run in isolated iframes, while permission manifests control sensitive capabilities.

Isolation does not make every extension harmless. The launcher itself supports shell execution, clipboard reads and writes, file access, scheduled work, and text replacement in other applications. Anyone deploying Asyar in a managed workplace should examine extension permissions and provider settings instead of treating it as a simple application finder. Optional encrypted sync also introduces a recovery phrase and passphrase-loss risk that the documentation explicitly leaves with the user.

Open issues show provider and Linux compatibility edges

GitHub recorded 523 stars, 5 combined open issues and pull requests, and a push on August 26, 2026. No latest GitHub release was returned by the API, even though the README points users to Releases for installers. The same-day push and issue updates show active work; the small combined count is not a count of confirmed bugs.

Issue 684 reports that a custom OpenAI-compatible provider rejected an Asyar request because the selected model did not accept the temperature field. Compatibility labels do not guarantee identical model parameters, so test the exact endpoint and model you intend to use. Issue 435 reports a WebKit process abort on Kubuntu 26.04 with an EGL error when launching the AppImage. One report does not define all Linux support, but it gives affected users a specific failure to check before migrating their daily launcher.

Choose it for breadth, after testing your desktop path

Asyar makes the most sense for someone who wants scripts, files, clipboard actions, extensions, and AI behind one keyboard shortcut. The source is GPL-3.0 for the core application, while the SDK and extension tools use MIT licensing. That split is clearly documented and relevant to anyone redistributing modified builds or creating extensions.

The 672 MB dependency footprint is manageable for desktop development, yet the failed build prevents a clean recommendation to contributors today. End users have easier installer paths, but Wayland limits, WebKit behavior, provider quirks, and extension permissions deserve a trial on the actual machine. If those checks pass, Asyar offers far more than application search. If they do not, a narrower launcher will be easier to trust and maintain.

Alternatives

ProjectWhat it isPick it when
UlauncherA Linux application launcher with shortcuts and community extensions.pick this instead when Linux is your only target and you want a launcher with less AI and data-handling scope.
Flow LauncherA Windows search and launcher utility with a plugin ecosystem.pick this instead when you use Windows and care more about mature launcher workflows than one cross-platform setup.
RaycastA polished macOS launcher with a large extension catalog and commercial services.pick this instead when macOS polish and an established store matter more than open source or Linux support.

What people are saying

  1. [github-trending] Xoshbin/asyar

Sources

  1. Asyar README
  2. Asyar repository
  3. Issue 684: incompatible temperature parameter
  4. Issue 435: Kubuntu startup crash

More dev tools reviews

workmux · v2rayNG · SecLists · hashcat · eslint · fastfetch · the whole board →