repository·main·Indexed 22 days ago
https://github.com/atomicbot-ai/atomic-chatAn OpenAI API-compatible local LLM interface and server that supports inference engines like llama.cpp and MLX. Documentation includes guides for the AutoQA E2E test runner, ReportPortal integration, and developer instructions for building extensions using the @janhq/core package.
autoqa/scripts/ directory contains platform-specific scripts used by the AutoQA GitHub Actions workflow. These scripts are designed to handle the lifecycle of Atomic Chat during automated testing, including downloading, installing, running tests, and cleaning up installations across Windows, Ubuntu, and macOS. Using these scripts improves workflow maintainability by moving complex logic out of YAML files and into dedicated, testable files.The apple-notes skill allows you to manage Apple Notes on macOS using the memo command-line utility. This enables creating, viewing, searching, editing, and exporting notes that sync across Apple devices via iCloud.
Requirements:
darwin) only.memo CLI installed via Homebrew.Notes.app.When to use:
When NOT to use:
obsidian instead).memory.notes.store instead).Jan provides several ways to interact with local AI, depending on your needs:
https://chat.jan.ai.The gog-workspace skill allows an agent to interact with Google Workspace services (Gmail, Calendar, Drive, Docs, Sheets, Contacts) using the gog CLI.
Key Requirements:
os.shell.run and skill.run_script.check-gog.sh via skill.run_script for diagnostics.gog operation must be an external CLI invocation using os.shell.run with cmd: "gog" and a separate args array. Do not use skill.run_script for general gog commands.darwin and linux.The documentation is organized into specific routes to help different types of users find information:
/guides/: Instructions for end users on how to use the Atomic Chat application./developer/: Documentation for developers looking to extend the software./api-reference/: Swagger/OpenAPI reference for the Jan API server./changelog/: Release history and changes./blog/: Project blog posts.Jan provides several advanced features beyond basic chat to help organize and extend your AI workflows:
Jan is available across multiple platforms to suit different workflows:
While desktop and server options are available, the project is prioritizing the web experience for browser-based usage.
The apple-calendar skill allows an agent to read Apple Calendar events using the icalBuddy CLI and create new events using AppleScript (osascript).
Requirements:
darwin).os.shell.run capability.icalBuddy must be installed via Homebrew, and the terminal/agent must have Calendar permissions in macOS Privacy & Security settings.Capabilities:
Limitations:
icalBuddy commands; all writes must use osascript.gog-workspace) or non-time-based tasks (use apple-reminders).Jan utilizes a family of specialized models optimized for autonomous agentic behavior (planning, executing, and completing tasks):
Deep Research is a methodology for generating comprehensive research reports by combining systematic web search with synthesis. It typically operates as a structured pipeline with distinct phases:
Key components include the use of 'thinking' models and the ability to use tools (like search engines and scrapers) to interact with the web.
In GPU programming, a kernel is a piece of CUDA code designed to execute a specific sequence of operations on the GPU. Kernels are launched within a thread block grid, which executes the kernel across multiple Streaming Multiprocessors (SMs) across the entire GPU.
Benchmarking is used to quantitatively compare kernel performance across different problem sizes and to understand how hyperparameters impact execution speed. This is critical for iteratively optimizing kernels to maximize GPU utilization.
ncu)