newest
Open in hackernews
Can I Use ChatGPT if I care about the climate?
https://charmindustrial.com/blog/can-i-use-chatgpt-if-i-care-about-the-climate
1
•
n2parko
•
2d ago
Anthony Scopatz on xonsh and shells in general
https://www.johndcook.com/blog/2015/11/15/xonsh/
1
•
Qem
•
1m ago
•
0 comments
NASA Tests Scalable Satellite Tech to Launch Sensors Quicker
https://www.nasa.gov/centers-and-facilities/langley/nasa-tests-scalable-satellite-tech-to-launch-sensors-quicker/
1
•
rbanffy
•
2m ago
•
0 comments
Fascism for First Time Founders
https://www.techdirt.com/2025/07/17/fascism-for-first-time-founders/
3
•
archagon
•
5m ago
•
0 comments
GitWear: Access GitHub on Your Watch
https://github.com/DaytimeDev/GitWrist
1
•
thunderbong
•
5m ago
•
0 comments
Make things that are easy to delete
https://medium.com/@jensenbox/make-things-that-are-easy-to-delete-55c4a513a813
2
•
jensenbox
•
9m ago
•
0 comments
Legendary Software Rendering Era with Sean Barrett [video]
https://www.youtube.com/watch?v=T1tXepGXDDM
1
•
todsacerdoti
•
9m ago
•
0 comments
3D object tracking using cheap cameras (example stealth planes)
https://github.com/ConsistentlyInconsistentYT/Pixeltovoxelprojector
1
•
punnerud
•
12m ago
•
0 comments
Prettier NPM Packages Compromised in Supply Chain Attack
https://socket.dev/blog/npm-phishing-campaign-leads-to-prettier-tooling-packages-compromise
1
•
feross
•
12m ago
•
0 comments
Type Checking with Symbolic Execution
https://bullno1.com/blog/type-checking-with-symbolic-execution
2
•
bullno1
•
14m ago
•
0 comments
Postgres to ClickHouse: Data Modeling Tips
https://clickhouse.com/blog/postgres-to-clickhouse-data-modeling-tips-v2
2
•
saisrirampur
•
18m ago
•
0 comments
Ask HN: What's the competitive advantage these days?
2
•
creepy
•
24m ago
•
2 comments
I built an MCP server for observability, here's my unhyped take
https://signoz.io/blog/unhyped-take-on-mcp-servers/
2
•
pranay01
•
26m ago
•
0 comments
Quantum Battery Has Outperformed a Classical One
https://www.popularmechanics.com/science/a65385727/quantum-battery-speed-limit/
1
•
Bluestein
•
27m ago
•
0 comments
The Golden Rule Goes Digital: Being Mean to LLMs Might Be Our Dumbest Gamble
https://substack.com/home/post/p-168721801
1
•
anonym29
•
28m ago
•
2 comments
Giving Up on Element and Matrix.org
https://xn--gckvb8fzb.com/giving-up-on-element-and-matrixorg/
21
•
upofadown
•
30m ago
•
6 comments
Show HN: KarmaWall – A daily public wall for small productivity wins
https://www.karmawall.space
1
•
SiddharthBasava
•
31m ago
•
0 comments
NPM Fund
https://www.gojiberries.io/npm-fund/
1
•
Brajeshwar
•
31m ago
•
0 comments
The underfunded mystery of sudden cardiac arrest
https://nymag.com/intelligencer/article/sudden-cardiac-arrest-death-medical-research-funding.html
1
•
bookofjoe
•
32m ago
•
1 comments
Show HN: DOMTranslator package to translate text nodes in DOM
https://github.com/translate-tools/domtranslator
1
•
vitonsky
•
33m ago
•
0 comments
BumpTop (2007) [video]
https://www.ted.com/talks/anand_agarawala_rethink_the_desktop_with_bumptop
1
•
notmysql_
•
34m ago
•
0 comments
For privacy and security, think twice before granting AI access to personal data
https://techcrunch.com/2025/07/19/for-privacy-and-security-think-twice-before-granting-ai-access-to-your-personal-data/
1
•
rntn
•
34m ago
•
0 comments
An Empirical Investigation into Moral Attitudes Against Market Transactions
https://link.springer.com/article/10.1007/s10551-025-06049-y
2
•
XzetaU8
•
37m ago
•
0 comments
Penny-sized laser could help driverless cars see the world so much clearer
https://www.livescience.com/technology/electric-vehicles/penny-sized-laser-could-help-driverless-cars-see-the-world-so-much-clearer
2
•
Bluestein
•
38m ago
•
0 comments
Show HN: Open-source project that use LLM as deception system
1
•
mariocandela
•
39m ago
•
0 comments
Can Tinygrad Win?
https://geohot.github.io//blog/jekyll/update/2025/07/06/can-tinygrad-win.html
2
•
abhaynayar
•
40m ago
•
0 comments
Eyeless Creatures Have Emerged from One Cave After Hiding in the Dark
https://www.popularmechanics.com/science/animals/a65383962/dragon-pseudoscorpion/
2
•
Bluestein
•
41m ago
•
0 comments
OpenAI Is Building an office productivity suite
https://www.computerworld.com/article/4021949/openai-goes-for-microsofts-jugular-its-office-productivity-suite.html
4
•
ishita159
•
42m ago
•
1 comments
The World After Wireheading
https://geohot.github.io//blog/jekyll/update/2025/07/18/the-world-after-wireheading.html
2
•
abhaynayar
•
43m ago
•
0 comments
Ask HN: Tell us about how you fine tune your LLM
1
•
mrburton
•
44m ago
•
0 comments
Rethinking CLI Interfaces for AI
https://www.notcheckmark.com/2025/07/rethinking-cli-interfaces-for-ai/
18
•
Bogdanp
•
44m ago
•
6 comments