zabouti / #ai

https://karpathy.ai/zero-to-hero.html

https://www.beaverapp.ai/

Academic research assistant that lives in Zotero. Search your entire library, chat with PDFs, discover new papers, and get sentence-level citations — all from one side panel.

https://community.obsidian.md/plugins/agent-client

Chat with Claude Code, Codex, Gemini CLI, and more via the Agent Client Protocol — right from your vault.

https://notes.ansonbiggs.com/youre-probably-using-agent-skills-wrong/

Skills have quickly become the most important for writing code with agents. Here is how I do it, and bad patterns I see from others.

https://www.reddit.com/r/ObsidianMD/comments/1uai1w2/karpathys_llm_wiki_setup/

https://blog.boxcars.ai/p/from-second-brain-to-shared-brain

The hottest idea in AI this week is something a German sociologist figured out with index cards in 1953.

https://openchamber.dev/

OpenChamber is an agentic development environment for AI coding across desktop, browser, phone, and VS Code. Watch agents work, review diffs, branch sessions, and keep the whole board visible.

https://github.com/tomash/bookmarklets

Exports the current chatgpt.com, gemini, etc. conversation as a Markdown file. Runs immediately on click — no button injection, instant download.

https://gist.github.com/mkaramuk/44a44d83178e632ec0dd1f02186d822c

Supposedly the secret self prompt of Gemini

https://www.washingtonpost.com/technology/2026/05/23/ai-industry-super-pacs-are-intervening-midterm-congressional-races/

Nida Allam, a 32-year-old Democrat, was fighting a close North Carolina primary when a barrage of ads swept in. Over five days in February, the blitz across TV, radio and social media painted the incumbent, Rep. Valerie Foushee, as a liberal crusader. “She is leading the fight to hold ICE accountable,” one ad said of Foushee, who co-chairs a Democratic House commission on artificial intelligence.

https://aphyr.com/posts/411-the-future-of-everything-is-lies-i-guess

Once trained, a model can be run again and again cheaply. This is called inference. Notice that many of the examples involve images, something that computers have trouble with still.

https://www.shloked.com/writing/claude-code-source-patterns

A breakdown of the most interesting engineering patterns in Claude Code's leaked source code: composable system prompts, runtime instruction injection, context compression, forking, prompt caching architecture, and more.

https://github.com/run-llama/liteparse

A fast, helpful, and open-source document parser. Contribute to run-llama/liteparse development by creating an account on GitHub.

https://www.youtube.com/watch?v=iYpHOxgZjUE

video

https://dev.to/onsen/apfel-the-free-ai-already-built-into-your-mac-1kll

Apfel: The Free AI Already Built Into Your Mac - Complete guide and honest review for 2026. Tagged with discuss, news, tech, ai.

https://github.com/ammaarreshi/gemma-chat

Local AI chat + coding agent for Apple Silicon, powered by Gemma 4 via MLX / Supports Ollama - ammaarreshi/gemma-chat

https://www.betonit.ai/p/how-ai-will-change-higher-education

I’ve spent much of my career arguing that the main function of education is not to teach useful skills (or even useless skills), but to certify students’ employability. By and large, the reason our customers are on campus is to credibly show, or “signal,” their intelligence, work ethic, and sheer conformity. While we like to picture education as job training, it is largely a passport to the real training that happens on the job. How will AI change this? To start, we should bet against any major disruption. Sheer conformity is one of the top traits students want to signal, and doing anything radically new signals… non-conformity. Few students will stop attending traditional colleges to “go” to virtual AI training centers. And since most majors are already barely job-related, few students will rethink their majors either.

https://mcp.so/explore?q=obsidian

A list of MCP Servers and Clients with keyword: obsidian.

https://geminicli.com/docs/cli/tutorials/skills-getting-started/

Look here for help creating skills.

https://ollama.com/settings

Get up and running with large language models.

https://github.com/addyosmani/agent-skills

Production-grade engineering skills for AI coding agents. - addyosmani/agent-skills

https://github.com/addyosmani/agent-skills/blob/main/docs/opencode-setup.md

Production-grade engineering skills for AI coding agents. - addyosmani/agent-skills

https://addyosmani.com/blog/agent-skills/

This is additional documentation for an Agent Skill that I should look at. This may help me understand the idea of "context": "A skill is a markdown file with frontmatter that gets injected into the agent’s context when the situation calls for it." "A skill is not reference documentation. It is not “everything you should know about testing.” It is a workflow: a sequence of steps the agent follows, with checkpoints that produce evidence, ending in a defined exit criterion."

https://www.reddit.com/r/ObsidianMD/comments/1t2bbg5/turn_your_obsidian_canvas_into_a_board_of_titles/

The mode I live in is Headline mode: every Canvas card collapses down to just its # H1, so the canvas becomes a map of titles instead of a wall of paragraphs. Clicking a card opens a side panel with the full content, editable inline. If I just want to read through a flow without accidentally typing into things, I flip on Read-only mode. I don't understand how the Agent creates notes, but this looks great.

https://thoughts.jock.pl/p/local-llm-35b-mac-mini-gemma-swap-production-2026?r=nr6w1&triedRedirect=true

Justin sent me this. This is a substantial difference between Ollama and llama.cpp: So what does --mmap do? Instead of loading the entire model file into RAM (which is what Ollama tried, and why it choked), llama.cpp memory-maps the file. The OS treats the model like a virtual address space backed by your SSD. And Gemma 4 sounds fantastic.

https://www.youtube.com/watch?v=0uZpuZQi7Zs&t=48s

In this video, we build a fully self-hosted coding agent powered by the 7B parameter Qwen 2.5 Coder model, running on a GPU instance in Lambda Cloud and serv...

https://docs.vllm.ai/en/latest/

justin says that people are using this in place of Ollama

https://www.youtube.com/watch?v=WQ6xcjB-tqU&t=223s

Helps me distinguish agents, skills and commands

https://skills.sh/kepano/obsidian-skills

Discover and install skills for AI agents.

https://github.com/hgus107/A-Long-Walk-of-AI

A narrative walk through 90 years of AI history, paper by paper. 66 chapters from Turing 1936 to Blackwell 2025. - hgus107/A-Long-Walk-of-AI