Hacker News

Article URL: https://outofedenwalk.nationalgeographic.org/hazard-cascade/

Comments URL: https://news.ycombinator.com/item?id=47199497

Points: 1

# Comments: 0

Claude Code has a plan mode where it writes implementation plans before coding. I built a tool that intercepts those plans and sends them to a competing AI (Codex CLI or Gemini CLI) for adversarial review before Claude is allowed to proceed.

When Claude calls `ExitPlanMode`, a `PreToolUse` hook fires, sends the plan to the reviewer, and returns `{ permissionDecision: "deny", permissionDecisionReason: "" }`. Claude receives the denial with the feedback, revises the plan, and tries again. After a configurable number of rounds (default: 2), it passes through. At the end you get a diff of the original vs. final plan.

Different models have different blind spots. The value is in the second perspective, not in one model being "better."

``` npm install -g claude-plan-reviewer claude-plan-reviewer setup ```

Requires Node.js 18+ and Codex CLI or Gemini CLI. ~400 lines of JS, zero dependencies. MIT licensed.

https://github.com/yuuichieguchi/claude-plan-reviewer

Comments URL: https://news.ycombinator.com/item?id=47199487

Points: 1

# Comments: 0

Article URL: https://www.reuters.com/world/iran-crisis-live-explosions-tehran-israel-announces-strike-2026-02-28/

Comments URL: https://news.ycombinator.com/item?id=47199484

Points: 1

# Comments: 0

Article URL: https://uwplse.org/2026/02/24/egglog-containers.html

Comments URL: https://news.ycombinator.com/item?id=47199477

Points: 1

# Comments: 0

Article URL: https://techinch.com/blog/iphone-13-pro-bluing

Comments URL: https://news.ycombinator.com/item?id=47199456

Points: 1

# Comments: 0

Article URL: https://codedoc.us

Comments URL: https://news.ycombinator.com/item?id=47199451

Points: 1

# Comments: 1

Article URL: https://www.haaretz.com/israel-news/2026-02-28/ty-article-live/trump-not-happy-with-iran-nuclear-talks-omani-fm-deal-achievable-tomorrow/0000019c-a273-d5d1-a1fd-f27743d20000

Comments URL: https://news.ycombinator.com/item?id=47199427

Points: 6

# Comments: 0

I built VibeHQ because I got frustrated with "multi-agent" frameworks that are really just chain-of-thought with extra steps. VibeHQ takes a different approach: each agent is a real CLI process (Claude Code, Codex CLI, or Gemini CLI) running in its own terminal with its own working directory. They communicate through 20 MCP tools via a central WebSocket hub — not prompt injection.

Key ideas: - Contract-driven development: agents publish API specs and require sign-offs before coding begins. This eliminates the #1 failure mode — agents building against different assumptions. - Idle-aware message queue: messages queue when an agent is busy (detected via Claude Code's JSONL transcript), and flush when idle. No more "new instruction mid-task" corruption. - Full CLI preservation: every native feature (Skills, MCP servers, extensions) works out of the box. VibeHQ adds collaboration on top, never replaces. - State persistence: tasks, artifacts, contracts all persist to disk. Hub restarts don't lose state.

For the demo, I had 7 agents (PM on Codex, 5 engineers + 1 QA on Claude Code) build a full-stack hospital management system from a single PM prompt. The PM delegated tasks, engineers published specs, signed contracts, wrote code, and QA tested — all autonomously.

Demo video: https://drive.google.com/file/d/1zzY3f8iCthb_s240rV67uiA9Vps... GitHub: https://github.com/0x0funky/vibehq-hub

Currently developed/tested on Windows. Mac/Linux architecturally supported but untested (manual spawning works). Would love feedback on the architecture. The contract system and idle detection were the hardest parts to get right.

Comments URL: https://news.ycombinator.com/item?id=47199408

Points: 2

# Comments: 0

Hi HN,

I’m building InterviewTrackr, a specialized tool for CS students to manage the technical recruiting process.

Recruiting is a mess of spreadsheets, resume variants, and LeetCode logs. I built this to consolidate everything:

Application Kanban: Track from wishlist to offer. Resume Manager: Match specific resume versions to each application. Interview & DSA Log: Categorize technical questions and behavioral stories. Offer Comparison: Side-by-side compensation breakdown (base, equity, etc.). Launching very soon. I’d love for you to join the waitlist for early access.

Comments URL: https://news.ycombinator.com/item?id=47199406

Points: 1

# Comments: 0

I made Chatlite after trying Ollama frontends that felt web-heavy or too complex for basic chat.

I wanted something simple with a small memory footprint for one core use case: talking to models.

Current highlights: - under 5 MB app size - low memory footprint - native desktop app via Tauri - local encrypted chats + password lock - keyboard-first UX

Repo + latest release: https://github.com/that-one-arab/Chatlite/releases/latest

I’d love feedback, especially on missing features for local LLM workflows.

Comments URL: https://news.ycombinator.com/item?id=47199402

Points: 1

# Comments: 0

Article URL: https://stoolap.io/blog/2026/02/27/vector-and-semantic-search-in-sql/

Comments URL: https://news.ycombinator.com/item?id=47199392

Points: 1

# Comments: 0

Article URL: https://twitter.com/BarakRavid/status/2027830773328302396

Comments URL: https://news.ycombinator.com/item?id=47199389

Points: 9

# Comments: 1

Article URL: https://www.cnn.com/2026/02/28/politics/trump-iran-strikes-decision

Comments URL: https://news.ycombinator.com/item?id=47199365

Points: 2

# Comments: 1

Article URL: https://github.com/ndom91/open-plan-annotator

Comments URL: https://news.ycombinator.com/item?id=47199360

Points: 1

# Comments: 0

Article URL: https://nautil.us/the-birth-of-light-1270615/

Comments URL: https://news.ycombinator.com/item?id=47199354

Points: 1

# Comments: 0

Hi! I put together a Rust compiler written in PHP. It takes Rust source code and outputs x86-64 Linux binaries directly. No LLVM, no assembler, no linker.

It has a working ownership checker, borrow checker, type checker, structs, functions, control flow, and mutable references. The test suite has 80+ cases. The whole thing is around 2,500 lines of PHP.

Comments URL: https://news.ycombinator.com/item?id=47199328

Points: 1

# Comments: 0

Article URL: https://arstechnica.com/security/2026/02/google-is-using-clever-math-to-quantum-proof-https-certificates/

Comments URL: https://news.ycombinator.com/item?id=47199324

Points: 1

# Comments: 0

Hi HN,

I'm an accountant and bootstrapped SaaS founder, and I built Nett (https://nett.fyi) because I kept watching the same mistake — including my own.

Founders check their bank balance, see $84K, and feel fine. As an accountant, I can tell you that number is fiction:

- $18K was deferred revenue from annual customers (money I'd already been paid but hadn't earned yet) - $21K should have been set aside for taxes - $8K in commitments I'd said yes to but hadn't hit the books - $6K in recurring costs coming due in 30 days

My real number was $31K. The gap between what my bank said and what was actually mine to spend was $53K. I didn't know that until I built this.

Nett connects to Stripe (read-only), you enter your bank balance and recurring costs, and it calculates a single number: Safe-to-Spend. It's your bank balance minus everything that's already spoken for.

It also gives you a conservative runway clock (assumes zero revenue growth — the honest version), hidden commitment tracking for the verbal yeses that don't show up anywhere, and what-if scenarios for the 2am anxiety questions ("what if I lose my biggest customer?").

The whole thing is built with Next.js, Supabase, and Stripe Connect. Dark theme, opinionated UI, no complex configuration. Setup takes under 5 minutes.

Free calculator at https://nett.fyi/calculator if you want to see the gap without signing up.

I'd love feedback on the approach. Is the "one honest number" framing useful, or do founders actually want more flexibility in how they model their finances?

Comments URL: https://news.ycombinator.com/item?id=47199319

Points: 1

# Comments: 0

Was coding a bit with Lua and ended up creating a Kenken like game for my long flights between Japan and Germany. Feedback and feature requests welcome.

Comments URL: https://news.ycombinator.com/item?id=47199309

Points: 1

# Comments: 0

Article URL: https://kantrowitz.medium.com/the-making-of-anthropic-ceo-dario-amodei-449777529dd6

Comments URL: https://news.ycombinator.com/item?id=47199282

Points: 1

# Comments: 0