Kevin Wu Won

692 posts

Kevin Wu Won banner
Kevin Wu Won

Kevin Wu Won

@kwuwon

just a vibe engineer

Sydney NSW, Australia เข้าร่วม Mart 2008
136 กำลังติดตาม96 ผู้ติดตาม
Kevin Wu Won
Kevin Wu Won@kwuwon·
@trq212 I tried it in my project with an existing CLAUDE.md (generated from the old /init) but it only made some cosmetic changes to it. I expected it to remove the things that were redundant. Is it better to delete the CLAUDE.md and start from nothing?
English
0
0
0
8
Kevin Wu Won
Kevin Wu Won@kwuwon·
My paperclips are on it
Kevin Wu Won tweet media
English
0
0
0
50
Kevin Wu Won
Kevin Wu Won@kwuwon·
Can someone please vibe code an autoresearch MiroFish to solve world peace?
English
1
0
1
39
Tanner Linsley
Tanner Linsley@tannerlinsley·
Ghostty was fun, but time for something else. I still love opencode, too but with CC plans dead on it… I’m feeling lost. Full GUI? T3 Code? Opencode GUI? Warp? Back to cursor? Try CC again? Raw Codex? My 🧠 hurts and I just need to keep shipping.
English
413
9
1.3K
331.1K
Kevin Wu Won
Kevin Wu Won@kwuwon·
After a surprisingly short interview, it found the essence of my philosophy
Kevin Wu Won tweet media
English
0
0
0
8
Kevin Wu Won
Kevin Wu Won@kwuwon·
Prompt of the day: "There are a lot of useless tests. Interview me in a binary-search way to figure out where I draw the "useless" line. Then write down guidelines in a skill with skill-creator"
English
1
0
0
24
Kevin Wu Won
Kevin Wu Won@kwuwon·
@trq212 Is this just one gigantic thread? Or can I tell it to find the right repo/thread and delegate to that? And create new repos/threads?
English
0
0
0
27
Thariq
Thariq@trq212·
We just released Claude Code channels, which allows you to control your Claude Code session through select MCPs, starting with Telegram and Discord. Use this to message Claude Code directly from your phone.
English
1.7K
2.4K
25.8K
7.4M
Boshen
Boshen@boshen_c·
Vite+: We'll be adding support for Bun as a package manager, but we are not planning on making bun work as a runtime at this time. github.com/voidzero-dev/v…
English
11
12
200
19.1K
Kevin Wu Won
Kevin Wu Won@kwuwon·
@owickstrom Writing code by hand?? How about a vibe coding UI optimised for eink? Where it's paginated so there's no scrolling
English
0
0
0
19
Kevin Wu Won
Kevin Wu Won@kwuwon·
@_sorrengailll It's always been terrible. This is why I'm switching back to Pixel when my iPhone's time is up
English
0
0
0
5
⚡︎
⚡︎@_sorrengailll·
yo what the FUCK did Apple do to the iPhone keyboard because I genuinely feel like I’m losing my mind typing these days
English
472
1.4K
24.3K
1.8M
euphoric.eth
euphoric.eth@StevenNevins·
@kwuwon @lydiahallie Yeah I think it’s basically a subagent with an SOP is what I can gather as the intent
English
1
0
0
26
Lydia Hallie ✨
Lydia Hallie ✨@lydiahallie·
Btw you can add `context: fork` to run a skill in an isolated subagent. The main context only sees the final result, not the intermediate tool calls It gets a fresh context window with CLAUDE.md + your skill as the prompt. The `agent` field even lets you set the subagent type!
Lydia Hallie ✨ tweet media
English
57
81
1.2K
134.8K
Kevin Wu Won
Kevin Wu Won@kwuwon·
@jtotheren @lydiahallie Came here to ask this. Seems like the main difference is that sub agents do not get the full prior context, only whatever main agent decides to tell it
English
1
0
0
15
euphoric.eth
euphoric.eth@StevenNevins·
@kwuwon @lydiahallie context: fork creates a isolated context currently. So the conversation history isnt shared between the main agent and skill agent
English
1
0
0
196
euphoric.eth
euphoric.eth@StevenNevins·
@lydiahallie the context:fork setting doesn’t make much sense imho given the behavior of /fork. I’d love an option for that to have the same behavior as /fork where a skill goes off an does some work with the shared history of the main agent and just returns the outcome to the main agent
English
4
0
7
2K
Kevin Wu Won
Kevin Wu Won@kwuwon·
@claudeai Good thing Claude is good at fixing the hardest problems
Kevin Wu Won tweet media
English
0
0
0
36
Kevin Wu Won
Kevin Wu Won@kwuwon·
There are three hard things in computer science: cache invalidation, naming things, off-by-one errors and running python. Why is it so hard? Here's @claudeai official skill-creator failing twice before using --break-system-packages. Did it just break my system? ¯\_(ツ)_/¯
Kevin Wu Won tweet media
English
1
0
1
61
Kevin Wu Won
Kevin Wu Won@kwuwon·
@jamonholmgren Try the superpowers brainstorming skill. You don't have to read the plan it produces because it interviews you about everything that goes into it, so you're "reviewing" it without realising
English
0
0
0
8
Kevin Wu Won
Kevin Wu Won@kwuwon·
@jamonholmgren Reading code and tests and manual QA is too laborious for humans. I don't want to have to click around and fake data to find out that the entire strategy was wrong. Often, it's something subtle that I wouldn't notice in a build, but I would pick up in an up front planning chat.
English
2
0
1
10
Jamon
Jamon@jamonholmgren·
To reiterate a few things so they don't get lost: 1. I never want to review another agent-produced plan again. Waste of my time, overwhelming, not worth it. It's valuable *to the agent*, but not to me. 2. I will burn all the tokens, run all the tests, do all the validations to make sure that when the work product lands on my desk, it's as good as the agents can make it. My time and energy is the most important thing here. 3. The feedback loop is critical: I'll work on the process, docs, and specs as much as I need to, in order to reap the benefits in future sessions. No more manual guidance via interactive sessions (with the exception of exploratory hacking).
Jamon@jamonholmgren

My current agentic workflow is about 5x faster, better quality, I understand the system better, and I’m having fun again. My previous workflows have left me exhausted, overwhelmed, and feeling out of touch with the systems I was building. They also degraded quality too much. This is way better. I’m not ready to describe in detail. It’s still evolving a bit. But I’ll give you a high level here. I call this the Night Shift workflow.

English
33
20
713
125K

ค้นพบ

@trq212 @amorriscode @tannerlinsley @boshen_c @cnakazawa @owickstrom @_sorrengailll @StevenNevins