Feed aggregator

Report from MPs warns of unknown number of unsafe subpostmasters convictions based on multiple pre-Horizon systems

Article URL: https://ngrislain.github.io/projects/2026-3-12-dont-vibe--prove/

Comments URL: https://news.ycombinator.com/item?id=47359103

Points: 1

# Comments: 0

Article URL: https://inceptive-ai.com

Comments URL: https://news.ycombinator.com/item?id=47359099

Points: 1

# Comments: 1

Article URL: https://www.nytimes.com/2026/03/12/technology/meta-avocado-ai-model-delayed.html

Comments URL: https://news.ycombinator.com/item?id=47359074

Points: 2

# Comments: 1

Article URL: https://www.reuters.com/sustainability/boards-policy-regulation/adobe-announces-ceo-transition-shares-fall-2026-03-12/

Comments URL: https://news.ycombinator.com/item?id=47359039

Points: 3

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=47359037

Points: 2

# Comments: 0

Article URL: https://www.algotradeai.live/

Comments URL: https://news.ycombinator.com/item?id=47359026

Points: 2

# Comments: 0

There have been lots of talk about PRDs being dead and spec-driven development etc.

Here is how we are planning. We iterate with our team and agents on one plan document.

It starts with some of the key things in a PRD but shorter and to the point: goals, use cases, jobs to be done, differentiators, strategy, succcess.

Then, in the same document, we include architecture diagrams in excalidraw or mermaid.

Then we describe the implementation approach but without any code, the approach.

Goal is as short as possible while being thorough... usually a few pages.

Now, we have one unified document that we/agents can use to build. The agents use this plus a full testing plan plus the code. And we get much better results.

It is nice because a human can and will read a few page document and the agents will keep it up-to-date with what is implemented.

Comments URL: https://news.ycombinator.com/item?id=47359022

Points: 2

# Comments: 0

Article URL: https://increment.com/software-architecture/in-space-no-one-can-hear-you-kernel-panic/

Comments URL: https://news.ycombinator.com/item?id=47359010

Points: 2

# Comments: 1

Article URL: https://rivian.com/newsroom/article/rivian-introduces-r2-lineup

Comments URL: https://news.ycombinator.com/item?id=47359004

Points: 3

# Comments: 0

How puffnfresh, my coworker, made a single command to download, assemble, and boot the VM image is arguably as cool as the port itself!

Comments URL: https://news.ycombinator.com/item?id=47358988

Points: 2

# Comments: 0

Article URL: https://syndicode.com/blog/csp-failure-rails/

Comments URL: https://news.ycombinator.com/item?id=47358986

Points: 2

# Comments: 0

Article URL: https://github.com/LvcidPsyche/auto-browser

Comments URL: https://news.ycombinator.com/item?id=47358970

Points: 1

# Comments: 0

Article URL: https://www.wsj.com/tech/ai/ai-bots-claude-openclaw-285ac816

Comments URL: https://news.ycombinator.com/item?id=47358963

Points: 1

# Comments: 1

Article URL: https://www.businessinsider.com/ai-compute-compensation-software-engineers-greg-brockman-2026-3

Comments URL: https://news.ycombinator.com/item?id=47358961

Points: 1

# Comments: 0

Relay Architecture is a new way to run AI coding agents. Instead of one agent executing your entire plan until it degrades, you run a relay. Each fresh agent reads the full plan, picks up a minimal handoff from the previous one, works at peak capacity, and passes the baton forward. The team covers a distance no single agent could sustain.

The problem it exists to solve: LLMs degrade as context fills up — this is measured, not anecdotal (Chroma Research). Every AI coding tool today runs a single agent start to finish. By phase 4 of your plan, half the context window is spent remembering what was already done. I was restarting sessions by hand, copy-pasting progress notes just to keep quality up. That's not autonomous development. That's babysitting. The architecture has four components: the Master Plan (your PRD, read fresh by every agent — never accumulates in context), the Baton (a hard-capped 40-line handoff note — the constraint is intentional, a bloated handoff recreates context rot in the next agent), the Signals (trigger phrases agents emit so the orchestrator dispatches without understanding the code), and the Context Budget (real-time token tracking, automatic handoff at threshold).

Tarvos is the reference implementation for Claude Code. Each session runs in its own git worktree. Accept merges to your branch, reject discards cleanly.

Open source, MIT. Works today, rough edges exist.

Comments URL: https://news.ycombinator.com/item?id=47358960

Points: 1

# Comments: 0

Article URL: https://www.osti.gov/biblio/3013883

Comments URL: https://news.ycombinator.com/item?id=47358958

Points: 1

# Comments: 0

I built Parevo Core because I kept rewriting the same auth + tenant + permission logic across Go projects.

Each time: JWT validation, refresh rotation, session revoke, tenant context, RBAC checks, SQL filters for tenant_id. Same patterns, different codebases.

So I extracted it into a modular library. You pick what you need: auth, tenant, permission, storage, cache, lock, billing, etc. Each module has interfaces; you plug in MySQL, Postgres, MongoDB, or Redis. No framework lock-in—works with net/http, chi, gin, echo, fiber, GraphQL.

What’s included:

Auth: JWT, OAuth2, SAML, LDAP, API keys, WebAuthn, magic link Multi-tenant: lifecycle, feature flags, SQL filters Permission: RBAC, ABAC, cached checks Storage adapters: memory, Postgres, MySQL, MongoDB, Redis Extras: cache, lock, health, job queue, billing, GDPR export I’d love feedback—especially on the module boundaries and what’s missing for real-world SaaS.

GitHub: https://github.com/parevo/core Docs: https://parevo.github.io/core/

Comments URL: https://news.ycombinator.com/item?id=47358943

Points: 1

# Comments: 0

Article URL: https://randomlabs.ai/blog/slate

Comments URL: https://news.ycombinator.com/item?id=47358930

Points: 2

# Comments: 0

Article URL: https://darkwebinformer.com/full-source-code-of-swedens-e-government-platform-leaked-from-compromised-cgi-sverige-infrastructure/

Comments URL: https://news.ycombinator.com/item?id=47358920

Points: 2

# Comments: 0