amical

Amical is a local-first AI dictation and note-taking application that runs on your machine. It leverages Whisper for speech-to-text, open-source LLMs for processing, and integrates with your environment to provide context-aware dictation and voice-driven automation. The MCP (Model Context Protocol) aspect is mentioned as an integration in the features, enabling voice commands to interact with supported apps, but the README does not provide a runnable MCP server entry point or configuration. If you’re using MCP in your workflow, you can treat Amical as a client-capable product that exposes voice-command capabilities within the app, allowing dictation, summaries, and structured notes, with potential extension points for command-based automation via the app’s integration surface.

To use the available capabilities, install Amical on your machine, launch the app, and follow the in-app prompts to enable local model setup and privacy-first configuration. You can start dictation, switch contexts to the active app, and use built-in hotkeys or voice commands to create notes, request summaries, or perform other workflow actions. The MCP-related feature is described as extensible and related to voice commands that can control your apps, so look for command surfaces or settings within Amical to enable or tailor these interactions to your environment.

Prerequisites:

• A supported operating system (e.g., macOS, Windows) with the ability to install applications.
• Homebrew installed on macOS or an equivalent package manager on your platform.
• Internet access for initial installation and model downloads.

Install via Homebrew (macOS):

brew install --cask amical

Alternatively, follow the official download paths for Windows or other platforms from the Download section of the README (links provided there).

Post-install steps:

• Launch Amical from your Applications menu (macOS) or Start menu (Windows).
• Allow initial model downloads and complete any privacy/configuration prompts.
• Enable MCP-related features or voice command integrations within the app settings if you plan to automate workflows.

If you need to rebuild or install from source, refer to the project’s docs for environment setup and build instructions specific to your platform.

Notes and tips:

• Amical emphasizes privacy and offline capability with local models; ensure your environment has enough resources for on-device processing.
• The MCP integration is mentioned as a capability for voice-driven automation, but the README does not provide a runnable server configuration. If you intend to use MCP, check the app settings for available command surfaces or integration hooks.
• For macOS users, the recommended installation path is via Homebrew as shown in the Download section. If you encounter issues, verify that your system meets the minimum requirements for local model inference (CPU/GPU, memory).
• Keep the app updated to access the latest privacy, security, and feature improvements related to dictation, context awareness, and any MCP-related capabilities.