Adam | Paragraph

adam-

Apr 20

Rodd Redwing photographed by Ralph Crane, circa 1956

adam-

Apr 19

Embracing the work

adam-

Apr 17

adam-

Apr 16

Who here has put mem0 to the test? https://mem0.ai/

Mem0 - The Memory Layer for your AI Apps

Mem0 enables AI apps to continuously learn from past user interactions, enhancing their intelligence and personalization.

adam-

Apr 16

After 13 years...we are so back! https://www.youtube.com/watch?v=6bghDcbzfEU

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

adam-

Apr 15

Shout out to @czar for creating the Farcaster Times Great way to keep up on whatever worthwhile conversations are happening on here without having to constantly be here. https://farcaster.xyz/miniapps/Ewhnx45ecbpv/the-farcaster-times

Farcaster Mini App

The Farcaster Times

Get the best of Farcaster delivered to your inbox

adam-

Apr 14

Not letting them off the hook here, but I've noticed that without seeing how calls are being rerouted, this kind of diminished outcome seems inevitable. Tools like this are going to become more popular & necessary as these kind of failures (both from a corporate response and model perspective) surface. I don't see how a mid/large size organization can properly incorporate LLMs without implementing this kind of layer in the mix. https://www.clawtrace.ai/

ClawTrace

ClawTrace is observability for OpenClaw agents. See failures, cost leaks, and the next fix to improve reliability.

adam-

Apr 14

This is the skill issue delta. You have to know how to effectively prompt to get results like this, which 99% of people are unaware of. They're stuck in treating prompts the same way they tackle a google search. For people here who have been experimenting and flexing their prompt game, the possibilities seem endless. You start looking for problems to solve and put them to the test. There's also a dopamine hit you get when you find a baseline prompt that can be applied to a broad surface area of your expertise.

adam-

Apr 14

Had a dream that Apple released a 32-inch MacBook, called the MacBook Pro Ultrawide, and it looked like this. I bought one and unlocked extreme productivity but then it wouldn't fit into my backpack, so I had to leave it behind.

adam-

Apr 13

adam-

Apr 12

Schroedinger's ceasefire

adam-

Apr 11

adam-

Apr 11

Tangentially related

adam-

Apr 10

In my quest to minimize token usage, but retain quality of results. Who else here is using caveman? https://github.com/JuliusBrussee/caveman

GitHub

GitHub - JuliusBrussee/caveman: 🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman

🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman - JuliusBrussee/caveman

adam-

Apr 10

Not sure how many people here experienced FARK dot com in the early days, but one thing they did that I've seen every successful community replicate since, is a weekly image contest. Any active community I'm still part has implemented this strategy to successfully sustain DAU and keep their most fervent members active during bust cycles. Alchemix has done a particularly good job at tying these contests into lead ups for community calls. Participants vote and the winner gets a nominal cash prize. As a result, turn out is consistently high. Farcaster solved for this in. way with the weekly leaderboards, which I know people want to see re-instated. I can't remember if @neynar scores were part of the ranking process as time went on, but it would seem like a good way to combat spam. https://www.marshu.com/collection-pictures-fark-photoshop-contests.php

Best Of Fark Photoshop Contests

best of fark photoshop contests

adam-

Apr 9

I've had two game ideas I've been iterating on for years. Today I finally put one of them in motion, fleshed out a proper outline and defined the mechanics. Goal is to create a proof of concept to show off the mechanics and gameplay. if it tracks I'll build it out fully. AI pipeline is Godot MCP + Filesystem MCP + GitHub MCP to assist with the build and AEsprite to create the characters Looking to see if anyone here has used the Godot MCP and can speak to its pros and cons. https://github.com/ee0pdt/Godot-MCP

GitHub

GitHub - ee0pdt/Godot-MCP: An MCP for Godot that lets you create and edit games in the Godot game engine with tools like Claude

An MCP for Godot that lets you create and edit games in the Godot game engine with tools like Claude - ee0pdt/Godot-MCP

adam-

Apr 8

I give it a few months before we see the thinking revealed in the Claude leak show up in open source models, and implemented more efficiently. By that I mean all the glaring API and token draining bugs that were surfaced are patched, while retaining the agent tree structure to deal with complex context windows. Until then, people have to watch this dog and pony show play out every few months as they run older models into the ground. https://x.com/noahepstein_/status/2041861090645049768?s=46