Arthur-Ficial/apfel
Apfel
The free AI already on your Mac. CLI tool, OpenAI-compatible server, and interactive chat — all on-device via Apple Intelligence. No API keys, no cloud, no downloads.
Features
- On-Device FoundationModel Bridges - Bridges CLI and server requests to Apple's on-device FoundationModel through private system frameworks.
- Apple Intelligence CLI Tools - Provides a command-line interface for querying Apple's on-device FoundationModel without cloud dependencies or API keys.
- Multi-turn Interaction Managers - Maintains conversation context across multiple exchanges with configurable trimming strategies and system prompts.
- MCP Server Connections - Connects Model Context Protocol servers so the on-device model can invoke external tools during generation.
- Local On-Device AI - Runs a local language model on a Mac for interactive conversations without cloud dependencies or API keys.
- Model Context Protocol Integrations - Connects external tool servers via MCP, enabling the on-device model to discover and invoke tools during generation.
- Offline Chat Clients - Provides an interactive chat application that runs Apple's 3B-parameter language model locally with multi-turn conversation support.
- MCP Protocol Integrations - Connects the on-device model to external tools through the Model Context Protocol for discovery and execution.
- CLI-Based Model Querying - Sends natural-language prompts to Apple's on-device FoundationModel from the terminal.
- Local LLM Execution - Runs Apple's on-device 3B-parameter language model as a UNIX command with JSON output.
- On-Device Model Runtimes - Runs Apple's on-device FoundationModel locally, passing prompts and printing responses to stdout.
- AI-Powered - Pipes macOS system data into the local model for real-time analysis, command generation, and clipboard manipulation.
- AI Chat REPLs - Ships an interactive read-eval-print loop for back-and-forth conversation with automatic context trimming.
- Live System Summaries - Pipes live macOS system data into the local AI model for real-time plain-English summaries.
- Interactive Chat Components - Manages multi-turn conversations with automatic context trimming and system prompt support, all running on-device.
- OpenAI-Compatible Servers - Hosts a local HTTP server implementing the OpenAI chat completions endpoint with streaming, tool calling, and CORS.
- OpenAI-Compatible API Servers - Hosts a local HTTP server implementing the OpenAI chat completions endpoint with streaming and tool calling.
- MCP Tool Attachments - Attaches MCP servers to give the on-device model access to calculators, APIs, and databases via tool calling.
- Guardrail Relaxation Mechanisms - Provides a flag to reduce Apple's safety filter strictness for creative or long prompts.
- Context Trimming Strategies - Maintains conversation history across exchanges with automatic trimming to fit within the model's token budget.
- Shell Command Translators - Converts plain-English requests into executable shell commands, one-liners, or git summaries.
- Model Response Streaming - Outputs the model's reply token by token as it is generated, showing results incrementally.
- Shell Command Generation - Generates shell commands from plain English descriptions for file operations and git workflows.
- System Prompts - Allows providing custom instructions or personas that guide the model's behavior for the entire session.
- Prompt - Reads file contents and includes them in the prompt sent to the model for context-aware processing.
- Clipboard Content Transformations - Reads, transforms, and writes back clipboard text for tasks like grammar correction, translation, and summarization.
- Stdin/Stdout Pipeline Models - Accepts input from stdin and outputs text or JSON to stdout, enabling UNIX-style piping and shell integration.
- AI-Powered System Analyses - Pipes macOS process, disk, network, and battery information into the model for real-time analysis and plain-English summaries.
