Scott's Dope Skills

As I mentioned in a recent issue of MacStories Weekly for Club members, I believe that reliable dictation and text-to-speech are largely solved problems in the AI industry right now for most languages. There are certainly subtle differences between the latest models and not-so-subtle discrepancies when you consider local (and free) transcription models versus cloud-hosted

A meaningful text can be hidden inside another, completely different yet still coherent and plausible, text of the same length. For example, a tweet containing a harsh political critique could be embedded in a tweet that celebrates the same political leader, or an ordinary product review could conceal a secret manuscript. This uncanny state of affairs is now possible thanks to Large Language Models, and in this paper we present Calgacus, a simple and efficient protocol to achieve it. We show that even modest 8-billion-parameter open-source LLMs are sufficient to obtain high-quality results, and a message as long as this abstract can be encoded and decoded locally on a laptop in seconds. The existence of such a protocol demonstrates a radical decoupling of text from authorial intent, further eroding trust in written communication, already shaken by the rise of LLM chatbots. We illustrate this with a concrete scenario: a company could covertly deploy an unfiltered LLM by encoding its answers within the compliant responses of a safe model. This possibility raises urgent questions for AI safety and challenges our understanding of what it means for a Large Language Model to know something.

Check out Oracle’s post on agents memory: https://fandf.co/4dd3HOw---Obsidian was always my favorite note taking system where I've been happily building and ...

Consciousness is not separate from the physical world — our “soul” is of the same nature as our body and any other phenomenon of the world.

Causal evidence that AI assistance reduces persistence and impairs unassisted performance across math reasoning and reading comprehension tasks.

Thesis: reliable agents tackling complex tasks need deterministic control flow encoded in software, not increasingly elaborate prompt chains If you’ve eve...

Local models need focus and polish.

Sorry to use a technical term in the title

AI coding agents take the shortest path to done, which usually means skipping the specs, tests, and reviews that make software reliable at scale. Agent Skill...

['Programmer', 'Computer nerd', 'Unix enthusiast'];

More than 65 years ago, complex clinical diagnostic reasoning cases were introduced as the gold standard for the evaluation of expert medical computing systems, a standard that has held ever since. In this study, we report the results of a physician ...

36 likes, 7 comments. "Pi Agent (Full Course)"

112 likes, 15 comments. "Pi Coding Agent Setup After 2 Months"

Pi extension for async subagent delegation with truncation, artifacts, and session sharing - nicobailon/pi-subagents

The most powerful AI software development platform with the industry-leading context engine.

Set up Gemma 4 running in LM Studio, connected to Pi as the terminal agent

The Pi Coding Agent is what Claude Code is not: open source, customizable to the core, and built for engineers who want full control of their agentic engineering stack. Learn why mid-to-senior engineers should use Claude Code AND Pi together for outsized results.

16 likes, 1 comment. "Meet DESIGN.md: A new open standard for AI-generated UI"

These are commands I use with agents, mostly Claude - villainscode/pi-agent

332 likes, 56 comments. "Qwen3.6 vs Gemma 4: Which Actually Remembers Your Code?"

79 likes, 25 comments. "How I Make My Stunning Slideshows in ClaudeCode"

Comparison between open source PI agent and closed source Claude Code agent - disler/pi-vs-claude-code

Engineers... Claude Code got cancer. 🔥 Growth, profit motives, and serving the masses over the original niche audience of mid-to-senior engineers. But inste...

I ran Gemma 4 as a local model in Codex CLI I wanted to know whether Gemma 4 could replace a cloud model for my day-to-day agentic coding. Not in theory, in practice. I use Codex CLI every day …

I tested 6 AI coding harnesses for autonomous execution. Honest takes on Claude Code, Codex CLI (GPT-5.4), Aider, OpenCode, Pi, and Cursor, with SWE-bench scores, real token costs, and the +16pt harness effect most comparisons ignore.

llmfit - Hundreds of models and providers. One command to find what runs on your hardware.