About
JavaScript/python developer with experience in full stack development and QA automation.
Activity
-
I recently released version 6.0.0 of my language, dclang. dclang is a small, highly efficient, and capable general-purpose language designed for…
I recently released version 6.0.0 of my language, dclang. dclang is a small, highly efficient, and capable general-purpose language designed for…
Liked by Steven Kveton
-
After a long consideration, I have decided to shut down DocuCodes. I've learned so much over the last six months working on DocuCodes full time (and…
After a long consideration, I have decided to shut down DocuCodes. I've learned so much over the last six months working on DocuCodes full time (and…
Liked by Steven Kveton
-
Around this time last month, we wrapped filming of '20 Zomers', a short film written/directed by Jos Rikze and produced by me (WOOF Productions). It…
Around this time last month, we wrapped filming of '20 Zomers', a short film written/directed by Jos Rikze and produced by me (WOOF Productions). It…
Liked by Steven Kveton
Experience
-
-
-
-
-
-
-
-
-
-
-
Education
-
University of Illinois Urbana-Champaign
-
-
Activities and Societies: -Phi Delta Theta -Illinois Club Tennis -Rho Epsilon
More activity by Steven
-
As someone who has somehow miraculously continued to thrive (so far) in the new era of tech (by fully embracing ai adoption and being fortunate…
As someone who has somehow miraculously continued to thrive (so far) in the new era of tech (by fully embracing ai adoption and being fortunate…
Liked by Steven Kveton
-
- Why do most programmers use dark theme when writing code!? - Because light attracts bugs! 🐞 🤣 🤣 🤣 #Codding #QA #SQA #IT #Programming
- Why do most programmers use dark theme when writing code!? - Because light attracts bugs! 🐞 🤣 🤣 🤣 #Codding #QA #SQA #IT #Programming
Liked by Steven Kveton
-
Excited to accept a promotion to Principal Software Engineer at AutoTune. 🥳 Can't wait to work on some of our upcoming projects this year and…
Excited to accept a promotion to Principal Software Engineer at AutoTune. 🥳 Can't wait to work on some of our upcoming projects this year and…
Liked by Steven Kveton
-
We’re proud to have been part of the 2026 Carat Innovation Summit 🚀 This year’s theme explored how brands can balance powerful technology with…
We’re proud to have been part of the 2026 Carat Innovation Summit 🚀 This year’s theme explored how brands can balance powerful technology with…
Liked by Steven Kveton
-
I quit my job! In all seriousness, I want to say thank you to the people of Legacy. Thank you for the opportunities to create kick ass work…
I quit my job! In all seriousness, I want to say thank you to the people of Legacy. Thank you for the opportunities to create kick ass work…
Liked by Steven Kveton
-
A month into the new year feels like a reasonable time to look back on 2025… right? Last year, I got to work on a bunch of projects I’m really…
A month into the new year feels like a reasonable time to look back on 2025… right? Last year, I got to work on a bunch of projects I’m really…
Liked by Steven Kveton
-
As 2025 came to an end, my time with Warmer unexpectedly reached its conclusion. While layoffs are never easy, especially so close to the holidays,…
As 2025 came to an end, my time with Warmer unexpectedly reached its conclusion. While layoffs are never easy, especially so close to the holidays,…
Liked by Steven Kveton
Other similar profiles
Explore more posts
-
Jaiydev Gupta
Anthropic released Agent Skills on Friday, and I've been seeing a lot of confusion about what they actually are. Some people think they're just workflows repackaged. They're not. Here's the core difference: Who invokes what. Workflows are user-invoked. Agent Skills are LLM-invoked. Think about how MCP already works. Resources are user-invoked, but tool calls are made by the LLM. Skills follow the same pattern but at a higher level of abstraction. Before Skills, we had a real problem. LLMs would get confused about when to invoke tools. Anyone who's used Claude Code or other AI coding editors has seen this. The workaround was writing workflows to enforce certain tool behaviors, which meant you were doing the orchestration work that the LLM should have been handling. Now with Agent Skills, you can define how to use a tool in a SKILL.md file. Take Anthropic's webapp-testing skill as a real example (https://lnkd.in/g93ZyeTt). It wraps around Playwright, the browser automation tool. The skill teaches Claude: When to use Playwright for testing web applications How to structure test cases What verification steps to take How to handle UI debugging The flow becomes: LLM sees you need to test a web app, invokes the webapp-testing skill, the skill shows which Playwright methods to use, and the LLM follows the defined testing pattern. This is the real power. Your LLM is now using an agent skill that has detailed instructions on tool usage, rather than trying to figure it out on its own every time. The opportunity here is to develop generic skills that wrap around your MCP tools or specific actions in your codebase. Make them reusable. Make them clear about when they should be invoked and how they should execute. That's where you extract the most value.
2 Comments
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content