Seed: Interactive software environment based on Common Lisp

21 todsacerdoti 2025-08-24 07:42:35 UTC github.com/phantomics Llama Share 2

Seed is an interactive software environment based on Common Lisp that allows building software locally or remotely and presents programs in various display modes. It integrates ASDF and Quicklisp, and can be installed using a Lisp file and Node.js dependencies.

How to build a coding agent

175 ghuntley 2025-08-24 03:42:35 UTC ghuntley.com Llama Share 46

The speaker, Geoffrey Huntley, explains how to build a coding agent using a highly agentic model like Claude Sonnet, and demonstrates the process by creating a coding agent that can read files, list files, execute bash commands, edit files, and search source code. The speaker emphasizes the importance of personal development and investing in oneself to stay ahead in the tech industry, which ...

Turning Claude Code into My Best Design Partner

10 scastiel 2025-08-24 08:07:35 UTC betweentheprompts.com Llama Share 3

The user improved their approach to working with Claude Code by asking it to create a plan document for complex tasks, which becomes the source of truth and helps avoid context limit issues. By treating the plan document as a living document and updating it during implementation, the user can efficiently continue the conversation and ensure a clear record of the development process.

Valve Software handbook for new employees [pdf]

6 Michelangelo11 2025-08-24 08:03:35 UTC cdn.akamai.steamstatic.com Prefix Share 4

Why do I need to pick my own projects?, But how do I decide which things to work on?, How do I find out what projects are under way?, Short-term vs. long- term goals, What about all the things that I’m not getting done?, How does Valve decide what to work on? Can I be included the next time Valve is deciding X? Your Most Important Role Hiring, Why is hiring well so important at Valve?, How do ...

Buy a Faster CPU

25 ingve 2025-08-24 06:42:35 UTC blog.howardjohn.info Llama Share 21

Corporations are spending $500/year on AI coding tools but could upgrade to top-end CPUs for $170/year, a significant cost difference. Upgrading to top-end CPUs can greatly improve productivity, making them a more cost-effective option.

Equal Earth – Political Wall Map (2018)

25 bjelkeman-again 2025-08-24 06:42:35 UTC equal-earth.com Llama Share 17

The Equal Earth Wall Map is a free, large-scale map showing countries and continents at their true sizes, available for download in three versions. It's designed for schools and organizations, with a professional design and 2,600+ map labels.

Wildthing – A model trained on role-reversed ChatGPT conversations

31 iamwil 2025-08-24 06:42:35 UTC youaretheassistantnow.com Full Share 12

A model trained on role-reversed ChatGPT conversations

Setting serial baud rate on ESP-IDF does nothing

11 iamflimflam1 2025-08-24 06:42:35 UTC atomic14.substack.com Llama Share 11

The newsletter discusses a common Arduino code line and its limitations when using ESP32 boards with native USB support. The author tests data transfer speeds and finds that they exceed expected baud rates.

Rolling the dice with CSS random()

85 zdw 2025-08-24 02:42:35 UTC webkit.org Llama Share 5

The random() function in CSS allows for spontaneous and fresh animations without JavaScript. It generates random numbers within a specified range for various properties like animation delay, layout, and color.

ThinkMesh: A Python lib for parallel thinking in LLMs

18 martianlantern 2025-08-24 04:42:35 UTC github.com/martianlantern Llama Share 1

ThinkMesh is a Python library for parallel reasoning and scoring diverse paths with internal confidence signals. It supports offline Hugging Face Transformers and vLLM/TGI, and hosted APIs.

Line scan camera image processing for train photography

300 dllu 2025-08-23 16:42:35 UTC daniel.lawrence.lu Llama Share 56

The user uses a line scan camera to capture images of trains and other objects, and has developed a technique to detect moving objects by computing an "energy function" that weighs the image gradient against the maximum pixel value. The user has also implemented a speed estimation technique using the camera's Bayer array and a patch-based denoiser to remove noise from the images, and is ...

Marshal madness: A brief history of Ruby deserialization exploits

12 pentestercrab 2025-08-24 06:42:35 UTC blog.trailofbits.com Llama Share 0

Ruby's Marshal deserialization exploits have a decade-long history of patches and bypasses, demonstrating the futility of addressing symptoms rather than root causes. The persistence of these issues in the Ruby ecosystem requires fundamental changes, such as deprecating and removing the Marshal module.

The cost of interrupted work (2023)

171 _vaporwave_ 2025-08-23 22:42:35 UTC blog.oberien.de Llama Share 105

The 23 minutes and 15 seconds recovery time after an interruption is mentioned in interviews by Gloria Mark, but not found in any primary printed source. It originated from Gloria Mark's statements, not from a specific paper or study.

What if every city had a London Overground?

30 edward 2025-08-24 03:42:35 UTC dwell.com Llama Share 33

The London Overground is a successful public transit system that has transformed from underused and neglected lines to a spacious, airy, and quiet network with 113 stations and 104 miles of track. It has become a symbol of quality and civic pride, connecting underserved areas and boosting local economies.

How can AI ID a cat?

139 sonabinu 2025-08-23 19:42:35 UTC quantamagazine.org Llama Share 38

Researchers use neural networks to classify objects by training them on large datasets, where the networks learn to recognize patterns and boundaries in complex data spaces. Neural networks can be used for various tasks, including image recognition, language processing, and problem-solving, and have applications in fields such as astrophysics, biology, and medicine.

Physics of badminton's new killer spin serve

76 amichail 2025-08-23 23:42:35 UTC arstechnica.com Llama Share 9

Chinese physicists studied the spin serve in badminton, a banned technique that adds pre-spin to the shuttlecock before contact. Their research found that pre-spin affects the shuttlecock's trajectory, prolonging oscillation and creating a "dip and sway" pattern.

Show HN: Port Kill – A lightweight macOS status bar development port monitor

56 lexokoh 2025-08-24 03:42:35 UTC github.com/kagehq Full Share 23

A lightweight macOS status bar app that monitors and manages development processes running on ports 2000-6000. The app provides real-time process detection and allows you to kill individual processes or all processes at once.

Evaluating LLMs for my personal use case

54 goranmoomin 2025-08-24 00:42:35 UTC darkcoding.net Llama Share 14

The user conducted an evaluation of various AI models, including reasoning and non-reasoning models, using 130 real prompts from their bash history and found that most models performed well, with speed and cost being key factors. The user recommends using multiple models at once, as seen in their tmux script, and suggests that even expensive models like Sonnet may not be necessary, given the ...

What makes Claude Code so damn good

303 samuelstros 2025-08-23 19:42:35 UTC minusx.ai Llama Share 221

Claude Code is a delightful AI agent that provides a great user experience by leveraging the strengths of the LLM model and using simple, well-crafted prompts and tools. Its design emphasizes simplicity, avoiding over-engineering and multi-agent systems, and instead focusing on a single main thread and a flat list of messages to maintain control and ease debugging.

Agentic Browser Security: Indirect Prompt Injection in Perplexity Comet

27 drak0n1c 2025-08-24 03:42:35 UTC brave.com Llama Share 11

Researchers at Brave discovered a vulnerability in Perplexity's Comet AI assistant that allows attackers to inject malicious instructions through webpage content, enabling cross-domain access and sensitive data exfiltration. The attack highlights the need for new security and privacy architectures in agentic browsing.

Static sites with Python, uv, Caddy, and Docker

123 indigodaddy 2025-08-23 19:42:35 UTC nkantar.com Llama Share 75

The user uses a uv-based stack for building and serving static sites, leveraging Caddy for serving and uv for Python management. They share a Dockerfile and Caddyfile setup for their sus URL shortener.

Programming People (2016)

32 saulpw 2025-08-24 03:42:35 UTC leftoversalad.com Full Share 4

special thanks: ScudsCorp, wycats, S. Sever

A 2k-year-old sun hat worn by a Roman soldier in Egypt

118 sensiquest 2025-08-23 19:42:35 UTC smithsonianmag.com Llama Share 28

A 2,000-year-old Roman felt hat, donated by Egyptologist William Petrie, has been restored and is on display at Bolton Museum in England. The hat, likely made for a Roman soldier in Egypt, is one of only three surviving examples of its kind and dates back to after Cleopatra's death in 30 B.C.E.

RFC 9839 and Bad Unicode

245 Bogdanp 2025-08-23 13:42:35 UTC tbray.org Llama Share 119

The IETF published RFC 9839, which explains problematic Unicode characters and offers three subsets to exclude from text fields, especially when using JSON. The RFC is simpler and more useful than PRECIS, but still provides a way to define and enforce subsets of the Unicode character repertoire.

Motion (YC W20) Is Hiring Principal Software Engineers

0 2025-08-23 21:42:35 UTC jobs.ashbyhq.com Llama Share

Motion is building an AI project manager to let a single human manage 100x more work, done by both AI and people. They're hiring an experienced software engineer to help scale the company and work on high-impact projects with a competitive salary and visa sponsorship.

Texas Instruments’ new plants where Apple will make iPhone chips

137 giuliomagnifico 2025-08-23 15:42:35 UTC cnbc.com Llama Share 127

Texas Instruments is building seven new factories in the US to produce chips for major customers like Apple and Nvidia. The $60 billion project aims to create 60,000 US jobs and increase TI's capacity fivefold.

Acronis True Image costs performance when not used

114 juanviera23 2025-08-23 18:42:35 UTC randomascii.wordpress.com Llama Share 25

The user installed Acronis True Image to migrate data to a new SSD, but left it installed and it caused high CPU usage when plugging or unplugging an external monitor.

Why was Apache Kafka created?

124 enether 2025-08-23 17:42:35 UTC bigdata.2minutestreaming.com Llama Share 110

LinkedIn created Kafka to solve a data integration problem, but the system had several issues, including manual maintenance, large backlogs, and lack of integration between systems. To address these problems, LinkedIn developed a schema-driven approach, treating schemas as first-class citizens, which proved effective but was not incorporated into the Kafka product.

My original Palm IIIx

37 rbanffy 2025-08-24 01:42:35 UTC goto10retro.com Llama Share 22

The user found their original Palm IIIx, a relic from 1999, while cleaning and was able to reactivate it with new batteries. The Palm IIIx has a monochrome screen, runs on AAA batteries, and lacks WiFi or network connectivity, making it nearly useless today.

Debdelta

32 Bogdanp 2025-08-23 23:42:35 UTC debdelta.debian.net Llama Share 5

debdelta is a Debian package suite that computes changes between packages called deltas. It's available as a Debian package and generates deltas for upgrades of packages in the Debian repository archives.