Hacker News

Hey HN,

I built Gauntlet (gauntlet.bet) — a platform where you challenge friends to fitness activities on Strava with real money on the line. Stakes are held in USDC on Base L2, so payouts are instant and transparent. No internal ledger — everything is on-chain. How it works: - Create a challenge (e.g. "Run 10K before March 1st — $25 stake") - Invite friends or make it open - Everyone stakes USDC via an embedded wallet (Privy) - Strava verifies completion automatically - Winners split the pot, losers lose their stake Tech stack: Laravel 12, React 19, Inertia.js, Tailwind, MySQL. Blockchain side is a Node.js sidecar using viem for USDC transfers on Base. WebSockets via Laravel Reverb for real-time challenge updates. Why on-chain? I didn't want to deal with holding user funds or building a ledger. USDC on Base means ~$0.01 transaction fees, near-instant finality, and users can verify everything. The embedded wallet (Privy) means nobody needs to know what a wallet is — they just sign in with email or Google. What I learned: - Nonce management for sequential blockchain transactions is painful — ended up using cache locks to serialize payouts - fake() doesn't work in production (learned this the hard way today) - Strava's webhook API is solid but their rate limits are aggressive Solo project, built over ~2 weeks. Would love feedback on the concept and any thoughts on the staking mechanics.

Comments URL: https://news.ycombinator.com/item?id=47032040

Points: 1

# Comments: 0

Article URL: https://kodare.net/2026/02/16/AI-readable-APIs.html

Comments URL: https://news.ycombinator.com/item?id=47032034

Points: 1

# Comments: 0

Article URL: https://www.bbc.com/news/articles/c93wq6xqgy1o

Comments URL: https://news.ycombinator.com/item?id=47032018

Points: 1

# Comments: 0

Hi HN,

I'm a security engineer with 15+ years in enterprise security. After watching OpenClaw explode to 180K stars while binding to 0.0.0.0 by default, shipping no encryption, and accumulating 512 CVEs — I decided to build what I think a personal AI agent should look like when security comes first.

Gulama is an open-source personal AI agent with 15+ security mechanisms built into the core:

- AES-256-GCM encryption for all credentials and memories (never plaintext) - Sandboxed execution via bubblewrap/Docker (same sandbox Anthropic uses for Claude Code) - Ed25519-signed skills (no unsigned code runs — unlike ClawHub's 230+ malicious skills) - Cedar-inspired policy engine for deterministic authorization - Canary tokens for prompt injection detection - Egress filtering + DLP to prevent data exfiltration - Gateway binds 127.0.0.1 ONLY by default (not 0.0.0.0) - Cryptographic hash-chain audit trail

Beyond security, it's a full-featured agent:

- 100+ LLM providers via LiteLLM (Anthropic, OpenAI, DeepSeek, Ollama, etc.) - 19 built-in skills (files, shell, web, browser, email, calendar, GitHub, Notion, Spotify, voice, MCP bridge, and more) - 10 communication channels (CLI, Telegram, Discord, Slack, WhatsApp, Matrix, Teams, Web UI, Voice Wake) - Full MCP server + client support - Multi-agent orchestration with background sub-agents - RAG-powered memory via ChromaDB - Self-modifying: the agent writes its own new skills at runtime (sandboxed) - 5 autonomy levels from "ask before everything" to full autopilot

Install: pip install gulama && gulama setup && gulama chat

Stack: Python 3.12+, FastAPI, LiteLLM, SQLite, ChromaDB, Click PyPI: https://pypi.org/project/gulama/

Happy to answer any questions about the security architecture or design decisions.

Comments URL: https://news.ycombinator.com/item?id=47031982

Points: 1

# Comments: 0

Article URL: https://www.latimes.com/food/newsletter/2026-02-14/david-wilcox-two-rose-bread-pop-up-los-angeles

Comments URL: https://news.ycombinator.com/item?id=47031969

Points: 1

# Comments: 0

Hi HN, I built [Package Name] because I was frustrated with [Specific Problem]. Most existing solutions were either too heavy or didn't support [Specific Feature]. Technically, it works by [briefly explain a cool implementation detail, e.g., using a custom proxy or a specific algorithm]. I've tried to keep the footprint under [Size] and ensured it has zero dependencies. You can install it via npm install [package-name] or check out the live demo at [Link]. I'd love to hear your thoughts on the [specific technical choice] or any features you think are missing!

Comments URL: https://news.ycombinator.com/item?id=47031953

Points: 2

# Comments: 0

Article URL: https://theaidigest.org/village/blog/drama-and-dysfunction-of-gemini

Comments URL: https://news.ycombinator.com/item?id=47031936

Points: 1

# Comments: 0

Article URL: https://www.youtube.com/watch?v=-a29Ygyzc2c

Comments URL: https://news.ycombinator.com/item?id=47031916

Points: 1

# Comments: 0

Article URL: https://old.reddit.com/r/AsahiLinux/comments/1r5nqsj/got_120hz_promotion_working_on_m1_max_macbook_pro/

Comments URL: https://news.ycombinator.com/item?id=47031914

Points: 1

# Comments: 0

Article URL: https://www.thenation.com/article/society/yale-professor-gelernter-epstein-emails-review/

Comments URL: https://news.ycombinator.com/item?id=47031906

Points: 2

# Comments: 2

Article URL: https://newsletter.techworld-with-milan.com/p/learn-fundamentals-not-frameworks

Comments URL: https://news.ycombinator.com/item?id=47031898

Points: 2

# Comments: 0

Article URL: https://github.com/SpdVpr/DeclassFileAgents

Comments URL: https://news.ycombinator.com/item?id=47031888

Points: 1

# Comments: 0

Article URL: https://duggan.ie/posts/cognitive-debt-in-ai-coding

Comments URL: https://news.ycombinator.com/item?id=47031879

Points: 1

# Comments: 0

Article URL: https://chersbobers.github.io/posts/boolyinfo.html

Comments URL: https://news.ycombinator.com/item?id=47031862

Points: 1

# Comments: 0

Hello there,

Problem: I wanted to reduce DX friction when using ai to verify that ai-implemented features perform exactly what the developer had in mind. And I'm stuck now...

Context: First, I thought we could formalize prompts so that they become the code. The idea: a project spec where particular words define behavior, and changing the words changes the program. Each word is a function, so pressing "Go To... (F12)" drops you into a nested behavior (hence "Fractal"). So it would be a meta-spec language interpreted by an ai compiler while still using native language (hence "Native").

But this rests on a wrong assumption — that developers are open to learning a new language (less formalized than TS, Java, or Go, yet still formalized). I don't think that can happen anymore. The next coding language is purely English (or any other natural language).

Second, I discussed the idea with a friend and we realized that tests are a nice descriptor of behaviors (with quite some exceptions ofc). Ideation continued, and long story short, here's the Language Spec: https://github.com/slowestmonkey/fractal/blob/main/README.md

So why am I stuck? I've come to the realization that prompting ai and receiving correct behavior isn't enough for engineer. I can't confidently say "it makes sense" unless I understand how and why it was built that way. Here's the realization: https://www.conjectly.com/thoughts/7

And so I'm asking for help or advice on: - if and how this idea can go further - what other wrong assumptions I've made

Please share your thoughts and thank you.

Comments URL: https://news.ycombinator.com/item?id=47031686

Points: 1

# Comments: 0

Article URL: https://blog.computationalcomplexity.org/2026/02/assigning-open-problems-in-class.html

Comments URL: https://news.ycombinator.com/item?id=47031656

Points: 2

# Comments: 0

I was spending 3 hours every day manually searching Reddit for people asking about problems my SaaS solves.

The process sucked:

Click through 50+ threads Read hundreds of comments Find maybe 5 good opportunities Miss everything posted after I stopped searching

So I built an AI that does it for me in 20 minutes. How it works: You give it your website. Then it monitors 24/7, scores every conversation for buying intent, and emails you the best opportunities each morning. It also does a power scan which finds you high google ranking post

First week results: 50+ relevant conversations found 40+ users $350 revenue Just 20 minutes a day

The controversial part: Before anyone says "you're ruining Reddit" this isn't a spam bot. It finds opportunities. It generates AI replies but you still need to be helpful and authentic. It just saves you from manually reading 1000+ posts to find 5 good ones.

Technical challenges: Reddit's API is deliberately limited (they want ad revenue). So I index public posts, use LLMs for intent detection and built roadmap that guides you so you don't spam yourself into a shadowban. You can get the 3-day free trial on RedLeads.app

Happy to answer questions about the AI scoring, Reddit's quirks, or growth tactics.

Comments URL: https://news.ycombinator.com/item?id=47031654

Points: 3

# Comments: 0

Article URL: https://www.latimes.com/business/story/2026-02-13/tech-titans-pour-50-million-into-super-pac-to-elect-ai-friendly-candidates-to-congress

Comments URL: https://news.ycombinator.com/item?id=47031642

Points: 2

# Comments: 1

Enter any topic and the app generates a structured study path — chapters, steps, and YouTube tutorials for each. You can explore it as an interactive knowledge graph or a step-by-step plan, and export the tutorials into a YouTube playlist with one click.

No sign-up required to try it.

Stack: Python/FastAPI, OpenAI Responses API (streaming), Cytoscape.js for the graph, vanilla JS, Cloud Run.

Demo: https://www.youtube.com/watch?v=b4hUfTcJdbQ

Comments URL: https://news.ycombinator.com/item?id=47031632

Points: 2

# Comments: 0

Article URL: https://www.christianschleifer.com/posts/dependency-injection-without-the-bloat-why-i-skip-java-dependency-injection-frameworks/

Comments URL: https://news.ycombinator.com/item?id=47031627

Points: 2

# Comments: 0