fp.
newest
Open in hackernews
Lamdera v1.5.0 is out Elm 0.19.3, faster Wire, safer virtual DOM and++
https://dashboard.lamdera.app/releases/v1-5-0
1
•
cekrem
•
1h ago
One dead LED killed a working TV
https://train-it.eu/blog/one-dead-led-killed-a-working-tv/
1
•
ibobev
•
30s ago
•
0 comments
Humman Attention Requires Human Effort
https://gavinleroy.com/writings/posts/human-attention-effort/
1
•
yorelnivag
•
42s ago
•
0 comments
Ask HN: Should something be done about paywalled articles?
1
•
thejamesbox
•
1m ago
•
0 comments
EnsoJS: A native-first microframework for building declarative Web Components
https://ensojs.dev/
1
•
ibobev
•
1m ago
•
0 comments
An AI-Native Workspace, How realistic is that? [video]
https://www.youtube.com/watch?v=STQCVBBTOn0
1
•
vanessa1211
•
1m ago
•
0 comments
Show HN: Peregrine – Integrated Terminal and File Browser
https://github.com/David-OConnor/peregrine
1
•
the__alchemist
•
2m ago
•
0 comments
Is the Music Business Collapsing?
https://www.honest-broker.com/p/is-the-music-business-collapsing
1
•
RickJWagner
•
3m ago
•
0 comments
Show HN: CMusee – C/C++ timeless debugger
https://veefuse.com/en/cmusee/
1
•
veexx103
•
3m ago
•
0 comments
Tiny Terminal: A fake terminal where kids type anything and get fun responses
https://github.com/madebydia/tiny-terminal
1
•
surprisetalk
•
5m ago
•
0 comments
Model is being retired. Are you ready to upgrade?
https://mahalsystems.ai/research/what-survives-a-model-swap/
1
•
jasondavis
•
5m ago
•
1 comments
Show HN: Shufler: 3D gaming harness around your coding harness and dev pipeline
https://github.com/jgrichardson/shufler
1
•
jgrichardson67
•
5m ago
•
0 comments
Tailscale SSH: Findings and Fixes
https://tailscale.com/blog/tailscale-ssh-findings-fixes
1
•
mfrw
•
6m ago
•
0 comments
Where They Stand – Nonpartisan Guide to Every House and Senate Race
https://wheretheystand.vote
1
•
zaidbourini
•
7m ago
•
1 comments
Two high-speed trains collide near Toledo, Spain
https://www.euronews.com/video/2026/10/28/spain-first-footage-reveals-wreckage-of-deadly-high-spe...
2
•
evilmonkey19
•
9m ago
•
2 comments
Ask HN: Why do Agents keep on forgetting stuff?
1
•
rishikeshs
•
10m ago
•
0 comments
AI Can Draw Blueprints Now
https://stephenjgallagher.substack.com/p/ai-can-draw-blueprints-now
1
•
SteveGallag
•
13m ago
•
0 comments
Rudra-JS, LLM-designed product recommendations for React stores
https://github.com/clivedsouza1010/rudra-js
1
•
clivedsouza8484
•
13m ago
•
0 comments
Claude Chalk, a mod that lets Claude Code colour its own replies
https://github.com/adamkdean/claude-chalk
1
•
imdsm
•
15m ago
•
0 comments
Show HN: Popkern – open-source CAD kernel with G2 corner blends, browser IDE
https://popkern.dev/
1
•
geedy
•
18m ago
•
0 comments
Show HN: ReplySlip – An open-source async task inbox for AI agents
https://github.com/klahrich/replyslip
1
•
good1andrew
•
19m ago
•
0 comments
I Don't Write Code Anymore. I Still Read It
https://oioannou.com/blog/i-still-read-the-code/
2
•
GTP
•
19m ago
•
0 comments
Sleep framework: A new way for LLMs to retain knowledge in continual learning
https://research.google/blog/the-sleep-framework-a-new-way-for-llms-to-retain-knowledge-in-contin...
1
•
acossta
•
20m ago
•
0 comments
The Cost of AI in Code: Goodbye, Fingerspitzengefühl?
https://architecturenotes.co/p/the-cost-of-ai-in-code-goodbye-fingerspitzengefu
3
•
stevefan1999
•
21m ago
•
1 comments
Show HN: Veqen – Catch a leaked AI API key by its spend, within the hour
https://veqen-app.com/
1
•
udemex10
•
21m ago
•
0 comments
Chainlit is archived and no longer maintained
https://docs.chainlit.io/get-started/project-status
1
•
tonio
•
22m ago
•
0 comments
NovaWire – HTTP/HTTPS debugging proxy
https://novawire.dev
1
•
info_vewello
•
27m ago
•
0 comments
Chameleon templates, modernized: chameleon-rs
https://mkennedy.codes/posts/chameleon-templates-modernized-chameleon-rs/
1
•
lumpa
•
28m ago
•
0 comments
Please Let the Internet Die
https://www.wired.com/story/let-the-internet-die/
4
•
voxleone
•
29m ago
•
1 comments
Let Claude watch any video
https://github.com/admjal/claude-watchall
3
•
admjal
•
29m ago
•
0 comments
A Measurement Study of Contextual Leakage in AI Chat Platforms
https://arxiv.org/abs/2610.37467
1
•
ersinesen
•
30m ago
•
1 comments
Load More