Feed aggregator

A golf course directory and navigation app powered by OpenStreetMap and the Overpass API. Browse golf clubs worldwide, explore course layouts, and find pin distances.

GitHub: https://github.com/refarer/navtee

Comments URL: https://news.ycombinator.com/item?id=47294312

Points: 1

# Comments: 0

Article URL: https://folio.stage.obvious.ai/obvious/reactscope

Comments URL: https://news.ycombinator.com/item?id=47294305

Points: 1

# Comments: 0

I wanted web search I could pipe. Every tool I found either had an interactive UI (ddgr, googler), required a single paid API, or returned HTML. I wanted something that composes: pipe to jq, feed to an agent, use in scripts, no friction.

So I built qry. It's a hub binary that routes queries to adapter binaries via stdin/stdout JSON. The hub knows nothing about search engines. Adapters know nothing about routing. You swap adapters in a config file.

Four adapters today, three require no API key:

- DDG Lite scraping

- Brave Search HTML scraping

- Exa AI via the public MCP endpoint

- Brave Search API (requires key)

``` $ qry "numpy latest version"

[{"title":"...","url":"...","snippet":"..."}]

```

Install:

```

mise use -g go:github.com/justestif/qry@latest

mise use -g go:github.com/justestif/qry/adapters/qry-adapter-ddg-scrape@latest

mise reshim ```

The adapter protocol is plain stdin/stdout JSON so adapters can be written in any language. Docs on building one are in the repo.

https://github.com/justEstif/qry

Comments URL: https://news.ycombinator.com/item?id=47294295

Points: 2

# Comments: 0

I built a small Python library called SafeAgent that protects real-world side effects when AI agents retry tool calls.

One issue we ran into while experimenting with agent workflows is that retries can trigger irreversible actions multiple times:

agent calls tool ↓ network timeout ↓ agent retries ↓ side effect happens twice

Examples:

• duplicate payment • duplicate email • duplicate ticket • duplicate trade

Most systems solve this ad-hoc using idempotency keys scattered around different services.

SafeAgent centralizes this into a small execution guard.

The idea is simple:

1. every tool execution gets a request_id 2. SafeAgent records the execution receipt 3. retries return the original receipt instead of running the side effect again

Example:

FIRST CALL REAL SIDE EFFECT: sending email

SECOND CALL WITH SAME request_id SafeAgent returns the original execution receipt (no second side effect)

The project is early but includes examples for:

• OpenAI tool calls • LangChain style tools • CrewAI actions

PyPI: https://pypi.org/project/safeagent-exec-guard/

GitHub: https://github.com/azender1/SafeAgent

Curious how other people are handling retry safety for agent side effects.

Comments URL: https://news.ycombinator.com/item?id=47294291

Points: 2

# Comments: 1

Article URL: https://nullbook.ai/

Comments URL: https://news.ycombinator.com/item?id=47294281

Points: 2

# Comments: 1

Here are the answers for The New York Times Mini Crossword for March 8.

Article URL: https://archerzhang.me/forcing-flash-attention-onto-a-tpu

Comments URL: https://news.ycombinator.com/item?id=47294271

Points: 4

# Comments: 0

Article URL: https://507movements.com/

Comments URL: https://news.ycombinator.com/item?id=47294270

Points: 3

# Comments: 0

Im getting really burnt out with my main project and trying to turn it into a real company. So tonight I wanted to build something fun but still possibly useful.

While making a baked potato (takes about 90 mins) I tried to vibe code a pipeline checker. I do some consulting and its constantly a challenge for young sellers. We will be on a call with the whole team and their boss is grilling them on deal qualification and they don't know how to answer. Its embarrassing for them and they aren't learning anything.

I ended up building this BANT checker and training it with some of my knowledge on the qualification methodology. For an hour and a half it works pretty well. And its all poop themed!

How can I add more poop elements?

Comments URL: https://news.ycombinator.com/item?id=47294250

Points: 1

# Comments: 0

Article URL: https://www.oldversion.com/

Comments URL: https://news.ycombinator.com/item?id=47294177

Points: 1

# Comments: 0

Article URL: https://code.chuanqisun.com/zero-sum-game/

Comments URL: https://news.ycombinator.com/item?id=47294168

Points: 1

# Comments: 0

Article URL: https://github.com/pqrs-org/Karabiner-Elements

Comments URL: https://news.ycombinator.com/item?id=47294154

Points: 1

# Comments: 0

Article URL: https://cloud.google.com/blog/topics/threat-intelligence/coruna-powerful-ios-exploit-kit

Comments URL: https://news.ycombinator.com/item?id=47294152

Points: 1

# Comments: 0

Article URL: https://anishkokamd.substack.com/p/how-vinay-prasad-came-to-washington

Comments URL: https://news.ycombinator.com/item?id=47294144

Points: 1

# Comments: 0

Article URL: https://www.nytimes.com/2026/03/07/us/politics/judge-kari-lake-voa-layoffs.html

Comments URL: https://news.ycombinator.com/item?id=47294139

Points: 1

# Comments: 0

Article URL: https://www.youtube.com/watch?v=kxfT4ZzG2l0

Comments URL: https://news.ycombinator.com/item?id=47294134

Points: 1

# Comments: 0

I’ve followed BetaList and similar directories for years, and I regularly read side-project posts here on Hacker News.

With the explosion of AI coding assistants, I expected a noticeable increase in useful small SaaS tools and practical software. But honestly, I’m not sure I see it.

HN users have always built impressive things. One example I remember: a developer who quickly built a simple website to help his family find COVID vaccine appointments, it cost about $50 to build. Stories like that show how many practical builders are here.

Yet even with AI making coding easier, the quantity (and quality) of useful tools doesn’t seem to have increased dramatically.

It reminds me of this line from Charles Bukowski:

"but as God said, crossing his legs, I see where I have made plenty of poets but not so very much poetry."

Sometimes it feels like a modern version might read:

"as VC-backed tech said, sitting on their laurels, I see we created a lot of AI hype and vibe-coding platforms but not so much useful software."

What do you think the real bottleneck is now? Ideas, distribution, taste, persistence; or something else?

Comments URL: https://news.ycombinator.com/item?id=47294126

Points: 1

# Comments: 0

I built Yumo.to because I wanted a map-first way to discover onsens in Japan without digging through booking sites.

It currently covers 19,652 onsens across all 47 prefectures, built by merging national datasets, prefecture and municipal open data, OSM, and water-chemistry sources into one searchable map.

You can filter by tattoo policy, prices, opening hours, access, and some water data, and each detail page links back to its sources. I’d especially appreciate feedback on search quality, missing data, and which filters or trip-planning features would make it more useful.

Comments URL: https://news.ycombinator.com/item?id=47294094

Points: 1

# Comments: 1

I'm a solo carpenter. I was paying $150+/month for field service software and hated it — too bloated, too expensive for a one-person operation. So I built FieldFlow. My rules:

$5/month flat. No per-user fees. Ever. Invoicing, scheduling, client management — the three things I actually need Every piece of feedback gets acted on fast. No ticket queue, no roadmap committee — just me

Jobber starts at $39. Tradify is $47/user. I want to build something that solo contractors can actually afford, and I'm willing to keep the price at $5 forever if that's what it takes. I use it myself daily. That's the only quality filter I trust. Brutal feedback welcome. https://fieldflow-nine.vercel.app/auth

Comments URL: https://news.ycombinator.com/item?id=47294092

Points: 1

# Comments: 0

Article URL: https://www.wsj.com/us-news/immigration-protests-noem-minneapolis-0b8bd496

Comments URL: https://news.ycombinator.com/item?id=47294082

Points: 1

# Comments: 1