OpenWand- An AI cowork app for your desktop, Press a key, pick an intent, and OpenWandreads the right context, then streams an answer without making you leave what you're doing. Local-first, voice in/out, bring your own model provider. - SunnyLich/OpenWand
No reviews yetBe the first to leave a review for OpenWand
Maker
📌
I built OpenWand because I kept copying text into an LLM chat, re-explaining the context, and pasting the answer back while working in other apps.
With OpenWand, a prompt can take as few as two key presses: one to open OpenWand and one to choose an action. Selecting or changing a context source adds a third. The answer streams into a small overlay. To follow up, press the hotkey again or open the full chat window for a longer back-and-forth.
It also includes: -Selected-text, clipboard, active-app, browser, document, and screen-snip context—each source can be selected with one keypress from the overlay -Rewrite-and-paste back into the original field -Hold-to-talk voice queries and direct dictation -Local or optional cloud TTS -Hosted providers and local OpenAI-compatible endpoints such as Ollama and LM Studio -MCP client and context-server support -A full chat window for longer conversations -Isolated Python add-ons -Full privacy*
*OpenWand has no project-operated backend or hosted storage layer. Settings, chats, and optional memory remain on your machine, while API keys are stored in the OS keychain. Prompts and selected context go directly to your configured provider or local endpoint. Privacy mode is enabled by default to warn about or redact detected sensitive information before sending. Your chosen provider still receives anything you explicitly send to it.
OpenWand uses your own model connection—you can configure a supported provider/API key or connect a local server such as Ollama or LM Studio. OpenWand allows ChatGPT / Codex subscription access through OAuth.
Packaged builds are available for Windows 10+, macOS 13+, and Linux X11. Linux Wayland support is still in progress, and the latest macOS releases need additional community testing.
There is no paid or “pro” version; the complete app is MIT-licensed.
I’m looking for feedback, bug reports, and general impressions from anyone willing to try OpenWand. Let me know what works well, what feels confusing, what breaks, or what you think could be improved.
Repo, demo GIFs, documentation, and packaged builds: OpenWand
Report
the hotkey-to-intent flow is such a smart way to skip the chat window entirely. love that it grabs the right context automatically instead of making you copy-paste anything.
Report
Maker
@remzidarbadyy9 Thanks, Remzi. Keeping context selection quick and user-controlled was a major design goal—I wanted convenience without silently sending unnecessary data.
Report
Picked an intent from a key combo and got a streaming answer without alt-tabbing away from my editor, which felt surprisingly seamless for a v0 launch.
Report
Maker
@dorukzerolr8su Thanks, Doruk — glad the workflow felt smooth. Removing that constant editor-to-chat context switching was the main reason I built Wisp.
I built OpenWand because I kept copying text into an LLM chat, re-explaining the context, and pasting the answer back while working in other apps.
With OpenWand, a prompt can take as few as two key presses: one to open OpenWand and one to choose an action. Selecting or changing a context source adds a third. The answer streams into a small overlay. To follow up, press the hotkey again or open the full chat window for a longer back-and-forth.
It also includes:
-Selected-text, clipboard, active-app, browser, document, and screen-snip context—each source can be selected with one keypress from the overlay
-Rewrite-and-paste back into the original field
-Hold-to-talk voice queries and direct dictation
-Local or optional cloud TTS
-Hosted providers and local OpenAI-compatible endpoints such as Ollama and LM Studio
-MCP client and context-server support
-A full chat window for longer conversations
-Isolated Python add-ons
-Full privacy*
*OpenWand has no project-operated backend or hosted storage layer. Settings, chats, and optional memory remain on your machine, while API keys are stored in the OS keychain. Prompts and selected context go directly to your configured provider or local endpoint. Privacy mode is enabled by default to warn about or redact detected sensitive information before sending. Your chosen provider still receives anything you explicitly send to it.
OpenWand uses your own model connection—you can configure a supported provider/API key or connect a local server such as Ollama or LM Studio. OpenWand allows ChatGPT / Codex subscription access through OAuth.
Packaged builds are available for Windows 10+, macOS 13+, and Linux X11. Linux Wayland support is still in progress, and the latest macOS releases need additional community testing.
There is no paid or “pro” version; the complete app is MIT-licensed.
I’m looking for feedback, bug reports, and general impressions from anyone willing to try OpenWand. Let me know what works well, what feels confusing, what breaks, or what you think could be improved.
Repo, demo GIFs, documentation, and packaged builds:
OpenWand
the hotkey-to-intent flow is such a smart way to skip the chat window entirely. love that it grabs the right context automatically instead of making you copy-paste anything.
@remzidarbadyy9 Thanks, Remzi. Keeping context selection quick and user-controlled was a major design goal—I wanted convenience without silently sending unnecessary data.
Picked an intent from a key combo and got a streaming answer without alt-tabbing away from my editor, which felt surprisingly seamless for a v0 launch.
@dorukzerolr8su Thanks, Doruk — glad the workflow felt smooth. Removing that constant editor-to-chat context switching was the main reason I built Wisp.