I've completed "Playground" - Day 8 - Advent of Code 2025 adventofcode.com/2025/day/8 #AdventOfCode Whee!

I've completed "Laboratories" - Day 7 - Advent of Code 2025 adventofcode.com/2025/day/7 #AdventOfCode Definitely the hardest so far.

I've completed "Trash Compactor" - Day 6 - Advent of Code 2025 #AdventOfCode adventofcode.com/2025/day/6

I've completed "Cafeteria" - Day 5 - Advent of Code 2025 adventofcode.com/2025/day/5 #AdventOfCode

Thankfully the optimization I implemented to get partA to work, also worked beautifully for partB.

I've completed "Printing Department" - Day 4 - Advent of Code 2025 #AdventOfCode adventofcode.com/2025/day/4

After day 3 part 2, was happy to find both parts of day 4 a bit easier.

I've completed "Lobby" - Day 3 - Advent of Code 2025 #AdventOfCode adventofcode.com/2025/day/3

I've completed "Gift Shop" - Day 2 - Advent of Code 2025 #AdventOfCode adventofcode.com/2025/day/2

I've completed "Secret Entrance" - Day 1 - Advent of Code 2025 #AdventOfCode adventofcode.com/2025/day/1

Improve RAG with a simple technique - Prompt Works Chunk windowing is an incredibly simple context engineering technique that can improve the quality of augmented replies while keeping your costs predictable.

Chunk windowing is an incredibly simple context engineering technique that can improve the quality of augmented replies while keeping your costs predictable.

blog.promptworks.io/improve-rag-...

I don't tend to re-use prompts often, but here's one that's handy:

Take a look through the changes on this branch. Summarize them. Highlight any issues.

`git diff origin/main...HEAD`

The key to healthy code is the same for our bodies -> exercise. How early and how often do you run your code?

Richard Sutton – Father of RL thinks LLMs are a dead end Watch now (66 mins) | LLMs aren’t Bitter-Lesson-pilled

I thoroughly enjoyed this long form podcast with a legend in the industry, Rich Sutton.

> Large language models are trying to get by without having a goal or a sense of better or worse.

www.dwarkesh.com/p/richard-su...

Finally tried Gemma 3. Impressive all around and really happy with its vision capabilities. This is the 12b model running locally with Ollama and LibreChat. I'm getting a zippy 32 tokens/s on my M3 Max. Who else is going to the AI Engineer World's fair next week?

The best agent tool is just an agent with all its autonomy removed.

It’s still a difficult world in which to be relentlessly data driven.

@anthropic.com's Claude Desktop almost always searches the web now. So annoying! A lot of the time, I’d like the answer based on information in the model's weights. Agents just wanna call tools! #AIUXFail

🗓️ feat: Add Special Variables for Prompts & Agents, Prompt UI Improvements by danny-avila · Pull Request #7123 · danny-avila/LibreChat Summary Closes #4712 Closes #5007 Added replaceSpecialVars to consistently process variables on frontend/backend, using dayjs for date handling Extended variable support in prompts and agent instr...

Whee! LibreChat now has special variable expansion in prompts/agent instructions to support things like `{current_date}` github.com/danny-avila/...

There’s an art to knowing which corners to cut.

See how my AI powered software development has evolved over the last 2 years.

blog.promptworks.io/ai-powered-s...

Intent detection is still difficult. I'm tired of always having to say "use artifacts".

Yesterday I explained what wardriving and a cantenna are to some college students. 🤣 You know - the things I did for fun when I was in college...

LLMs mimic humans too well. When I give agents autonomy to write more code on their own they tend to make the same mistakes. They over engineer solutions before we prove there's any real value for the software to offer.

I put together a list of my favorite tools and services with brief descriptions. Are any of these in your daily use toolkit? What are your favorites that aren't on my list?

blog.promptworks.io/my-favorite-...

I've been enjoying the new agent mode in Cursor. There's definitely something here beyond just vibe coding. Not sure exactly what to call it. "Pairing With A Robot" maybe?

Manage your local notes and/or config file git repo with ease with one of my favorite new use cases for Local No-Code MCP Agents - The Home Directory Git Repo Agent.

blog.promptworks.io/local-no-cod...

I just used my first MCP server in @anthropic.com Claude MacOS App. This is going to be fun!

The cloudinary video link isn’t working well. Here it is on Vimeo. vimeo.com/1060330011/8...

Personalize the web in real-time with @anthropic.com Claude, @superwhisper.bsky.social, and loveable.dev

Remix the 10th mountain division hut information in less than 10 minutes!

Video: bit.ly/hut-data-remix

final result: bit.ly/loveable-huts

Most ideas that were outside the box 20 years ago are inside the box today.

Taxonomies are full of exceptions and contradictions.