frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Show HN: Self-contained offline knowledge cards with ULID-DNA and IDsEd25519

https://stoutenburger.com
1•tomneijman•1h ago
I built an open format for self-contained knowledge cards.

Each card is a single directory: - standalone HTML (no CDN, no external fonts) - bundled CSS + assets - print-ready PDF - JSON-LD metadata

No account. No platform. No external dependencies. The card itself is the product.

Goal: use LLMs to distill practical knowledge into small independent units — then make those units survive without the technology that created them.

"The best use of AI is not automation, but preservation — helping humans rebuild when machines are gone."

---

Identifier model:

Each card receives a 128-bit ULID. The bits are mapped to A/C/G/T, producing a 64-character DNA-like sequence.

The first 9 characters map to three craft “codon” words from a fixed vocabulary (e.g. klei·vuur·rots).

Properties: - Globally unique - Chronologically sortable - Bidirectionally decodable - No central registry

The DNA appears in the URL, footer, and print version.

---

Integrity:

Each build generates a SHA-256 manifest of all cards, signed with Ed25519. Verification happens client-side using the Web Crypto API (no server round-trip).

Verify here: https://stoutenburger.com/verify/

---

Edition 1 is live (10 cards): Water purification, fire, clay, basic metallurgy.

Five card types: Knowledge (why), Instruction (how), Product (what), Maker (who), Network (connections).

Build pipeline: Markdown + YAML → JSON Schema → Nunjucks → Puppeteer → signed manifest.

Cards are designed to roam: USB stick, any domain, or printed in a drawer.

The format is open. https://stoutenburger.com

Show HN: A minimal Claude Code clone written in Rust

https://github.com/keon/mini-claude-code
1•kwk236•39s ago•0 comments

Typing Fun Game

https://kwerty.site
1•soexya•1m ago•0 comments

Smartphones, Online Music Streaming, and Traffic Fatalities

https://www.nber.org/papers/w34866
1•speckx•1m ago•0 comments

Secria Mobile – Post-quantum encrypted email, now on iOS and Android

https://secria.me
1•adrianmav•2m ago•1 comments

Show HN: 20x – Open-source agent orchestrator for Linear/HubSpot tasks

https://github.com/peakflo/20x
1•dmitryv•3m ago•0 comments

Nobody Trusts Anybody Anymore

https://yourbrainonmoney.substack.com/p/low-trust-society-cost
1•lemonberry•3m ago•0 comments

Tell HN: YouTube Subscription page is being deprecated

1•csours•4m ago•0 comments

Any Open Source Contributors?

https://github.com/Shantanugupta43/SuggestPilot
1•shaanuknow•6m ago•1 comments

Show HN: Ask a graph database questions in plain English (runs in the browser)

https://cogdb.io/playground
1•am3141•7m ago•0 comments

Half of the side projects from HN make less than $1000 a month (2024-2025)

https://indielaunches.com/indie-maker-analytics-2024-2025-projects/
1•s-stude•8m ago•2 comments

Palo Alto's $400M scanner labels 91% of confirmed OpenClaw threats safe

https://oathe.ai/engineering/we-audited-1620-ai-agent-skills/
5•riyadaftary•8m ago•2 comments

Show HN: OnGarde – Runtime content security proxy for self-hosted AI agents

1•antimaterial•9m ago•0 comments

Show HN: SAIA – SCUMM for AI Agents

https://github.com/serendip-ml/llm-saia/blob/main/docs/INTRO.md
1•serendip-ml•10m ago•1 comments

SF Vibe House

https://www.sfvibehouse.com/
2•sahin•11m ago•0 comments

Palm OS User Interface Guidelines [pdf, 2003]

https://cs.uml.edu/~fredm/courses/91.308-spr05/files/palmdocs/uiguidelines.pdf
3•spiffytech•12m ago•0 comments

Show HN: SkillFortify, a formal verification for AI agent skills

https://github.com/varun369/skillfortify
1•varunpratap369•12m ago•2 comments

Microbial system to convert CO₂ into eco-friendly butanol

https://phys.org/news/2026-02-microbial-eco-friendly-butanol.html
1•PaulHoule•12m ago•0 comments

How to Close a Diamond Mine in the Northwest Territories

https://thewalrus.ca/how-to-close-a-diamond-mine-in-the-northwest-territories/
2•pseudolus•13m ago•0 comments

Anthropic is both too dangerous to allow and essential to national security

https://www.theargumentmag.com/p/anthropic-is-somehow-both-too-dangerous
1•ctoth•13m ago•0 comments

Kitty and Bitwarden Integration

https://github.com/pcaro/kitty_rbw
1•pcaro•14m ago•0 comments

Nvidia and Analyslop

https://www.wheresyoured.at/on-nvidia-and-analyslop/
1•7777777phil•16m ago•0 comments

I made a new AI disorder

https://www.generativemania.com/
1•darknoodle•18m ago•0 comments

Eustress and Distress: Neither Good nor Bad, but Rather the Same?

https://onlinelibrary.wiley.com/doi/10.1002/bies.201900238
1•amadeuspagel•20m ago•0 comments

TLA+ by Example

https://tlabyexample.com/
1•birdculture•21m ago•0 comments

AI=B+

https://www.publicbooks.org/aib/
2•geox•21m ago•0 comments

Child-free 'Disney adults' are transforming the company's theme parks

https://www.businessinsider.com/why-disney-parks-top-destinations-millennials-gen-z-2023-9
2•lordleft•22m ago•0 comments

Nvidia Linux Driver fork with P2PDMA support enabled on non-SoC platforms

https://github.com/us4useu/nvidia-open-gpu-kernel-modules
2•milaaaaaaa•23m ago•0 comments

The Model, the Chat and the Application

https://slowtechred.substack.com/p/the-model-the-chat-and-the-application
1•siscia•23m ago•0 comments

Anatomy of a Production AI Agent

https://www.mendral.com/blog/anatomy-of-a-production-ai-agent
1•shad42•24m ago•0 comments

Building a Pythonic REST client that feels like an ORM

https://blog.gofigr.io/posts/building-a-pythonic-rest-client
1•maciej_pacula•24m ago•1 comments