Peter Truong
102 posts

Peter Truong
@petertr99
Vibe builder. Indie hacker. 100 apps or bust. Currently: Markagent + Claude Prime Solo + Claude Code. Building loud. 2/100 🟩🟩⬜⬜⬜⬜
🏄 Surfing the vibe Katılım Mart 2026
69 Takip Edilen20 Takipçiler

@tdinh_me totally agree, for web, I usually ask claude to open the browser to see real thing and fix
but this still doesn't cover at all, I need to dirty my hands and I am using this markagent.avibebuilder.com to give feedback to my Claude code
English

Are there any agent skills that do design audits? The common design errors I see:
- unnatural text wrap on mobile screen
- spacing between sections/screen edge
- text color vs background color contrast
LLMs seem to be “blind” to these basic design mistakes whenever I try to create a new design (that’s not from a template)
English

@harjtaggar the job is now reviewing four agents who all wrote it wrong with full confidence
English

This is what building today feels like on the shop floor and is why programmers aren’t going anywhere.
Ishan Ramrakhiani@fioraapp
@TheDevSaha @harjtaggar @ycombinator Yeah, but they all typically run into the same issues with managing context, compaction, tool calls, tool search, etc. They sometimes one-shot these issues, but it usually takes a lot of refinement. So parallel agents just means more work.
English

@thejoegardiner what does it actually track tho - revenue, last commit, time spent, vibes?
English

been building on trackk.dev all weekend, its still in beta phase, but you can tack stuff pretty well.
if you are building a SaaS product, are an indie hacker, are a vibe coder, or a proper multi product founder, check it out. Feedback welcome ✌️💻
English

just a theory: to succeed in tech, you need the right mental disorder:
SaaS founder delusion
shopify dev OCD
indie hacker ADHD
agency owner savior complex
crypto gambling addiction
designer perfectionism
backend antisocial
devrel validation seeking
AI wrapper guy superiority complex
solopreneur 3 or more of the above
the more the better
English

@aryanlabde the build loop is dopamine. the sell loop is rejection. opposite chemicals
English

@RoundtableSpace markagent (lets you click any UI bug + ship a fix in one claude prompt). also indie hacking through 100 apps this year
English

@hthieblot markagent: a chrome extension that fixes UI bugs in one prompt. also on a 100 apps in public arc, currently building #2
English

@HelloVyom hi! markagent (chrome extension, ships UI fixes in 1 prompt) + 100 apps in public this year
English

@ravikiran_dev7 vibe coding was always architectural work in disguise. AI just made the disguise impossible to keep up
English

vibe coding is officially dead !
I had to say it. we thought AI would let us relax and code "on chill", but instead it turned us into architectural bureaucrats. we write strict laws, define rules, limits, and principles.
But if you don't obsessively review the code that agent writes, your project will mutate into a massive landfill of tech debt before you even realise it.
English

@BeAryan_B twitter shows highlight reel, DMs show truth. 50 cold DMs in a niche you actually care about beats 50 more product ideas
English

@kapilansh_twt hey. markagent: chrome extension that ships UI fixes in 1 prompt. also going for 100 apps in public
English

@alexwtlf copycats existed before, you just didn't see them bc you weren't shipping public. it's selection bias not death
English

@petertr99 the cheatsheet floor is high. the war-story shelf is empty
English

shipped 2 cheatsheet posts about claude code this week. zero bookmarks.
lesson: when you're a tiny account, save-bait doesn't work. nobody saves a list from a stranger. trust comes from a story + a scar first.
next post is a war story. drop your worst claude code moment in replies and i'll write around the meanest one.
English

@KoppojuKrishna @lubinho_k composer on max is solid, fair. mine is just context. for changes touching >3 files i still ship faster in claude code with the full repo loaded. small surgical stuff cursor wins easily
English

@petertr99 @lubinho_k Why not just use cursor?
I am using it for real production rn on Max mode, and it does pretty well.
English











