Easy macOS and iOS Tips and Fixes
Kept is a local-first tool that archives AI chat conversations from multiple LLM providers as plain markdown files stored on your filesystem, enabling full-text search, semantic recall, and integration with tools like Obsidian. It supports manual and automatic capture, offers a knowledge graph and topic clustering, and includes a MCP server for integration with other clients, while maintaining user data privacy by default without any cloud dependency.
Shortcuts Playground is a free, open-source plugin for Claude Code and Codex that enables users to create Apple Shortcuts using natural language prompts, producing ready-to-import shortcut files on Mac. Developed over six months by Federico Viticci, it uses extensive documentation, validation loops, and Apple's own command-line tools to generate mostly accurate shortcuts, allowing users to automate tasks without advanced Shortcuts knowledge. The project aims to democratize and simplify automation on Apple platforms while anticipating future native AI-based solutions from Apple itself.
https://www.macstories.net/stories/introducing-shortcuts-playground/
Atomic is an open-source, self-hosted knowledge graph app that organizes notes, articles, and web clips by automatically embedding, tagging, and linking them into a connected knowledge base. It features semantic search, AI-generated wiki synthesis with inline citations, agentic chat for AI interaction scoped to user notes, and a visual spatial canvas to explore the relationships between ideas. Available across desktop, server, iOS, and browser extensions, Atomic helps users build an AI-augmented knowledge system that grows and organizes itself over time.
Tolaria is an open-source, free note-taking app designed for the AI era that organizes notes as Markdown files with YAML frontmatter, supporting native relationships, Git integration, and Claude Code features. Created by Luca, an experienced content creator and author, Tolaria offers block-based editing, rich Markdown output, version control, and seamless AI tool integration, catering to users who want a powerful, transparent, and flexible knowledge management system.
OpenAI’s updated Codex app for Mac introduces a native computer-use feature that interacts with multiple Mac apps in the background via advanced accessibility APIs, enabling precise control without bringing them to the foreground. This technology, evolved from the acquired Sky app, allows Codex to automate complex tasks more efficiently and accurately than other models by reading app interface hierarchies rather than relying on screen recording or AppleScript, marking a significant advancement in AI-driven Mac automation.
The article explores “abliterated” local large language models (LLMs) that have their safety guardrails—implemented via reinforced learning from human feedback—removed through a mathematical process called orthogonalization. Unlike traditional uncensored models, abliterated LLMs bypass refusal behaviors and respond directly to any prompt, offering a more raw, unfiltered conversational experience, though sometimes at the cost of stability and reasoning performance. These models appeal to users seeking unrestricted AI interactions on local machines without editorial constraints, highlighting a trade-off between openness and reliability in AI usage.
https://www.makeuseof.com/tried-abliterated-local-llm-nothing-like-others/