ihar

Every AI code reviewer has the same flaw: it guesses, then states the guess as fact. It'll flag a vulnerability in src/auth.ts:84, a file that doesn't even exist. I'm building Skein: code analysis tool where the AI is structurally incapable of making that up. How it works 🧵:

@ctatedev bookmarked! I’d add skein.build as a deterministic map to keep track what ur loop is building right now

The best agent loops need the right tools → agent-browser.dev Verify changes in a real browser → portless.sh No port conflicts. Worktree-friendly. → emulate.dev Emulate third-party APIs → ai-cli.dev Image + video gen via CLI

@tomosman opening up waitlist now: skein.build

@tomosman this is sick imo half the loop is just re-figuring out what the app already does though "go over every feature + behavior" isn't agent work, it's a map which is why i'm building a tool that gives agent loops deterministic ground truth to run on

This "loop" automation is nuts inside of Codex. "/goal go over every single feature in this app create a user story with expected behaviour based on the code keep a single canonical spreadsheet tracking the features status - when done switch loop to testing every user story and documenting all errors - when done fix every logistical error or ux error - test every user behaviour again post fix" Shoutout to @MatthewBerman for the heads up. Hundreds of user stories being worked through like it's nothing.

If 'AI that can't lie about your code' sounds useful, I'm opening waitlist for early access. skein.build I'll be building this in the open. Follow along.

It'll run as a GitHub App: on every PR it maps what your change really does and posts the impact, every claim cited. No commits to your repo. Public and private.

Every AI code reviewer has the same flaw: it guesses, then states the guess as fact. It'll flag a vulnerability in src/auth.ts:84, a file that doesn't even exist. I'm building Skein: code analysis tool where the AI is structurally incapable of making that up. How it works 🧵:

@ionleu Yep, thanks!

@ihar_dev Done. Got the invite in your dm?

hi everyone i've created a group for developers & engineers and we are 1000+ strong 🔥 if you are interested drop a 👋 and I'll send you the invite link

@ln_dev7 @shadcn @linear Looks great! Found typo in package.json Also consider updating to next.js v15.2.4+ to fix middleware bypass, so builders avoid vulnerable version

Open-source template for your management projects. GitHub : github.com/ln-dev7/circle Built with @shadcn and @linear 's design.

@bekacru looks solid 🚀

👀

@arseniys_ @pontusab ~5 people. We've tried different approaches of project management. Now we try to separate BE/FE and micro modules for each task, e.g. "update service X in file Y for BE, then implement UI component in folder X, write tests" and only then connect everything in one module

@ihar_dev @pontusab Thanks for that! What does your process look like in terms of planning out modules so that they're realistically lockable and one dev can complete their task with only the given access scope? How big is your team btw?

A monorepo makes even more sense with Cursor

@yashgtech00 classic

what did people use before git?

@florinpop17 would be awesome to use a tool like @zerodotemail with scheduled batch emails processing some day

Which tools are indie hackers using to send newsletters? I want to start a weekly newsletter about my journey. What I’m learning. What I’ve been doing, etc.

@arseniys_ @pontusab Strict rules for project commitments? E.g. locked folders/files for each task. Robust task management Tbh in my team we still struggle with conflicts in updates, but one-task-per-time with subtasks separation works for us

@pontusab How do you collaborate on a monorepo? I've been working with a team and it's hard to manage multiple ppl working on the same thing. We'll have one person comment smth out and another uncomment it in one day. If they negotiate all the time, when are they gonna lock in & work?