VprompOpen Vpromp

Vpromp / Vpromp for Windows is in development

Vpromp for Windows is in development

The desktop preview brings Vpromp’s name, teal colors, and V-shaped assistant to a small floating window. Its purpose is to help people communicate clearly with AI while keeping relevant project context close at hand. A public installer is not yet available from this page.

A small assistant with a simple input

The current prototype starts as a small V with eyes and a floating dot. You can drag it around the desktop and click it to open a compact message panel. The panel provides a place to describe what you need, see a response, or stop waiting for an ongoing request.

After session-key setup, the bot can send text requests to Gemini for writing, explanation, analysis, and planning. It does not silently read your desktop or send all your project files. Opening the larger workspace is a separate action for selecting and reviewing project context.

Build a brief grounded in your project

In the workspace, choose a project folder and select relevant text files. The preview looks for framework hints in supported manifests and lets you inspect selected content. You can add your goal and paste an earlier AI response when you need help correcting a misunderstanding.

The local brief builder arranges this material using MIND: Map facts, clarify Intent, examine Nuance, and Distill a useful request. This local step is a structured template, not an LLM call or a claim that the app has understood the whole codebase. Optional Gemini refinement is a separate operation requiring approval to send the selected information.

Task: Explain why this sign-in flow fails while preserving the current provider.

Useful context: the relevant route, the expected callback, and the error message.

Boundary: explain or propose changes; the current preview does not apply them to project files.

What stays under your control

The local project reader limits the number and size of selected files and excludes common generated directories, hidden files, and secret filenames. It also attempts to redact common credential patterns. Those checks are not a guarantee that every private value will be found, so review the exact excerpts before sharing them.

The Gemini key is kept in memory for the current app session rather than saved as project content. Provider requests may incur charges on the user’s own Gemini account. Stopping a request prevents the app from using its later response; it cannot recall information already transmitted to the provider.

Current capabilities and release limits

The development version supports local briefs, optional Gemini refinement, pasted-response follow-ups, editable output, Markdown export, and copying text to another AI conversation. Opening a supported provider website does not automatically submit the brief. The user reviews and sends the final text.

Autonomous file editing, command execution, browser control, and VS Code integration are not connected in this version. It also has no automatic updater. Local development installers are unsigned and have not been presented here as a public production release. Until a public release is available, use the web prompt builder and the examples linked below to improve your AI requests.

Try Vpromp