frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Show HN: Mcpsnoop – Wireshark for MCP (transparent proxy and live TUI)

https://github.com/kerlenton/mcpsnoop
44•kerlenton•7h ago•13 comments

Show HN: Pixel a Go-Go – GameBoy Camera and retro screenshot upscaling for iOS

https://pixel.cameron.software/
2•CameronBanga•42m ago•0 comments

Show HN: I'm trying to map all vacation rentals you can book direct

https://staythere.ai/
3•wthayer•50m ago•0 comments

Show HN: ContextCodeCache in Rust

https://github.com/colwill/ccc
8•colwont•2h ago•0 comments

Show HN: Bramble – Local-first password manager

https://github.com/flythenimbus/bramble
107•MegagramEnjoyer•1d ago•25 comments

Show HN: Topics, Not Feeds

https://blogsreader.com
8•jdeibele•4h ago•3 comments

Show HN: Scout.kids – Agents that find activies for families

https://scout.kids
3•davmar•1h ago•0 comments

Show HN: ctx – Search the coding agent history already on your machine

https://github.com/ctxrs/ctx
64•luca-ctx•1d ago•37 comments

Show HN: CLI that helps AI agents avoid vulnerable dependencies

https://github.com/clidey/deptrust
18•modelorona•2d ago•4 comments

Show HN: Opbox – CRDT based sync for text files on disk

https://www.opbox.dev/
8•sensodine•5h ago•2 comments

Show HN: Reading Assistant Physical Books Meta RayBans

2•roshangill•1h ago•0 comments

Show HN: Kontext – Move an AI chat's full context to another AI in one click

https://github.com/anuragmerndev/kontext-ai
5•anuragmerndev•4h ago•1 comments

Show HN: Durable AI agents without the workflow engine

https://www.noworkflows.dev/
4•iacguy•2h ago•0 comments

Show HN: Theta-spec harness agnostic config surface

https://github.com/tamarillo-ai/theta-spec
4•ivanbelenky•2h ago•0 comments

Show HN: Pulse v0.2.0

3•xerrs•2h ago•1 comments

Show HN: Dockside – I turned unused space around the macOS Dock into a workspace

https://hachipoo.com/dockside-app
5•prajwalshetty•5h ago•2 comments

Show HN: Mlx-serve – LLM inference server for Apple Silicon, written in Zig

https://mlxserve.com/
2•ddalcu•2h ago•0 comments

Show HN: Maestro – scaffold Go microservices and keep them in sync

https://github.com/Zagforge-Org/maestro
2•anzedev•2h ago•0 comments

Show HN: StillUpTo – Keep track of the side projects if they are still alive

https://stillup.to/
3•hakanu•4h ago•0 comments

Show HN: How clanker are you? A reverse Turing test

https://howclankerareyou.com/
5•niklio•3h ago•3 comments

Show HN: Pieces – Social network for people

https://try.piecesof.me/
58•domo__knows•2d ago•56 comments

Show HN: Sharing LLM Artifacts with your team

https://github.com/danielpang/dropway
2•d_pang•3h ago•0 comments

Show HN: Origin Myth Map-creation myths of civilizations on a world map

https://originmap.sunnyguha.com/
4•sol0invictus•7h ago•0 comments

Show HN: Pull Claude Code transcripts into your Codex session, and vice versa

https://contextify.sh/
5•bredren•6h ago•1 comments

Show HN: An AI-friendly CLI for public space data

https://www.npmjs.com/package/spacedata
2•oscarjpicazo•4h ago•0 comments

Show HN: Static Site Generator

https://github.com/MTBBK/OSMAN
2•MTBBK•4h ago•0 comments

Show HN: zkGolf – Competitive optimization of formally verified circuits

https://zk.golf/
69•rot256•1d ago•11 comments

Show HN: Auto-continue Claude Fable 5 the second your 5-hour limit lifts

https://github.com/wavever/CCLimitPing
2•wavever•4h ago•0 comments

Show HN: OM Core – multidimensional models without spreadsheet cell formulas

https://github.com/cloudcell/om-core
13•cloudcell•3d ago•3 comments

Show HN: Inkwell – An RSS reader for e-ink devices

https://kendal.codeberg.page/inkwell/
70•imkendal•1d ago•10 comments
Open in hackernews

Show HN: How clanker are you? A reverse Turing test

https://howclankerareyou.com/
5•niklio•3h ago
You write 8 text completions and open models score how predictable each word was too them. Predictable => clanker. You can share results with your friends.

The scoring checks every word you write against the model's logprobs. Right now I'm using Llama3.1, Deepseek v3 and Qwen3 to keep costs low. I tried to calibrate it so other models (chatgpt/claude) score 100% and interesting human responses score in the 10-30% range.

Totally free, no signup

Comments

niklio•2h ago
The metric used is per-word surprisal: -logprob of each word you type. This is just the same thing as per-word cross entropy or KL-divergence where the user distribution is one-hot. Calibrating it so text generated by frontier models scored poorly was a challenge at first. Originally ChatGPT was scoring around 54%. I'm still having trouble assigning high scores to the personalized Gemini and ChatGPT responses when I'm logged in because all my personal context gives surprising responses.

And yes, gibberish responses score very human :)

TheJCDenton•1h ago
Funny little game, would be even funnier to have a system to roast the prose of a friend on social media or even a screenshot
niklio•58m ago
Thanks! That's a great idea - i'll top up my fable budget and get started :)