I keep forgetting how open source projects work. "How did they handle auth?" "What pattern is this?" "How do the pros structure this?" So I built Source Dive a companion app for learning from popular codebases. One night. Claude Code + Codex. Here's what it does 🧵

We’ve participated in over 30 contests, earning $150K+ in rewards Every lead, finding, and thought is documented in organized threads — a game-changing mental map for tackling any new codebases

I’ve been building an AI agent for finding vulnerabilities in smart contracts. It has found several confirmed bugs in audit contests, including code4rena contests which codebases are pre-scanned with @zellic_io's v12 LLM auditor. Tuning the prompts of an AI agent to find

Discord chose React Native over fully native development and honestly, it wasn't about avoiding native code. It was about velocity. Here's why it worked: The problem: Maintaining separate Swift and Kotlin codebases became a bottleneck. Features shipped at different times on iOS

Prompting is a developer skill, not magic. ✨ The Prompt‑Driven Development series shows how to turn prompts into real outcomes: refactoring existing codebases, fixing a chat app, querying docs with MCP, building responsive UIs, improving documentation, and generating tests.

How do we make codebases that are agent ready? At @aiDotEngineer, our CTO @EnoReyes breaks down why agents need tight verification loops to succeed, and why most codebases don’t provide those signals yet. Teams that invest in agent readiness will see 5–10× returns.

Blackbox Agent end-to-end encrypted through CLI lets you run tasks on your private codebases or also run datascience tasks on datasets that are private to your company.

In 2026, we’re focused on enterprise AI that delivers in production codebases. We’re developing a robust on-premise AI coding solution to meet the large-scale security and operational requirements of our clients. See what’s coming up for us:

Converts codebases into prompts with token counting https://t.co/ruvSxVKncb

Since Codex is taking forever for larger codebases, I split my tasks across to 5-8 agents. the important part is that it's split by domain so you don't overwrite things. But Codex is pretty good at recognizing others' work and stops or asks if it should proceed. I've also

📈 Progress check: 7 PRs merged ✅ 3 projects contributed to Learning how real-world codebases work #OpenSource #DevLife #BuildInPublic #TechTwitter #LearningInPublic #pr

Meta and Harvard Researchers Introduce the Confucius Code Agent (CCA): A Software Engineering Agent that can Operate at Large-Scale Codebases Confucius Code Agent from Meta and Harvard shows how much performance on real world software tasks comes from scaffolding rather than

I’ve been thinking a lot about how to evaluate coding agents, especially when large codebases are involved. I wrote up how I’m approaching benchmarks and evaluation infrastructure, and why I think most existing benchmarks fall short. https://t.co/F9sfC2nqXX

I'm addicted to Claude code, it seems. Lately, I've developed this weird habit of long, uninterrupted sessions just watching Claude reason about code, execute it, and work its magic. I'll spend 3-4 hours with Claude as it plans tasks, navigates complex systems and codebases, and

We did a lowkey launch of the Conductor extension for @geminicli, but it's already the most popular (stars) extension we built ourselves ( https://t.co/7Wp537KLQd). It's quite a helpful way to plan and implement features for new or existing codebases. https://t.co/k8Q3hHY8Hq

built something fun, try it in your swift codebases! https://t.co/YrmN4VBlal

The pace is getting silly. Anthropic just open-sourced the code-simplifier agent they use internally on the Claude Code team Just ask Claude to use `code-simplifier` at the end of a long coding session... → it literally cleans up messy codebases + simplifies complex PRs 🤯

We tracked AI code contribution across our products. The newer the product, the more AI wrote the code. HackerRank App (launched Dec): 75% SkillUp (launched 2024): 39.5% Screen & Interview (our first products): 20-21% New projects start AI-first. Older codebases are catching

Context switching costs you 20 minutes every time. We just fixed it. Introducing Workspaces on Renard🦊 Manage multiple products, separate teams, and distinct codebases—without the mental overhead. Isolate your Context. Switch instantly. Outfox the chaos Dropping soon