Hacker News

Article URL: https://bmarti44.substack.com/p/i-put-a-full-jvm-inside-a-browser

Comments URL: https://news.ycombinator.com/item?id=47228960

Points: 3

# Comments: 0

Article URL: https://github.com/ivan-digital/qwen3-asr-swift

Comments URL: https://news.ycombinator.com/item?id=47228957

Points: 2

# Comments: 1

Article URL: https://www.cbc.ca/news/canada/british-columbia/time-change-british-columbia-9.7112139

Comments URL: https://news.ycombinator.com/item?id=47228952

Points: 2

# Comments: 0

Article URL: https://www.tomshardware.com/software/operating-systems/california-introduces-age-verification-law

Comments URL: https://news.ycombinator.com/item?id=47228947

Points: 4

# Comments: 1

Article URL: https://lookingatcomputer.substack.com/p/rare-not-random

Comments URL: https://news.ycombinator.com/item?id=47228931

Points: 2

# Comments: 0

Article URL: https://the-infrastructure-mindset.ghost.io/brilliant-developer-unreliable-operator/

Comments URL: https://news.ycombinator.com/item?id=47228379

Points: 1

# Comments: 0

Article URL: https://insideclimatenews.org/news/01032026/iowa-county-data-center-ordinance/

Comments URL: https://news.ycombinator.com/item?id=47228371

Points: 1

# Comments: 0

Article URL: https://haptics.lochie.me/

Comments URL: https://news.ycombinator.com/item?id=47228358

Points: 1

# Comments: 0

Article URL: https://en.wikipedia.org/wiki/Mary_Mallon

Comments URL: https://news.ycombinator.com/item?id=47228349

Points: 2

# Comments: 0

Article URL: https://github.com/ivanbelenky/nbdantic

Comments URL: https://news.ycombinator.com/item?id=47228340

Points: 1

# Comments: 0

I've been using Claude Code heavily and kept wondering "why did this session cost $2?" or "why did it read the same file 15 times?".

Argus is a VSCode extension that analyzes Claude Code sessions from the .claude directory and gives you:

- Step-by-step breakdown of what Claude did - Cost analysis (which tools burned the most tokens) - Performance insights (retry loops, duplicate reads, context pressure) - Token usage visualization (cache hits, compaction events) - Flow diagrams showing file dependencies

Extension: https://marketplace.visualstudio.com/items?itemName=argus-cl...

It's basically a "time machine debugger" for your Claude Code sessions. You can click through each step, see tool inputs/outputs, and understand exactly what went wrong (or right).

Tech stack: TypeScript, React 19, Chart.js, Vite

Some interesting findings from my own sessions: - 40% of cost was from reading the same files repeatedly - Retry loops cost ~$0.15 per loop on average - Cache hit ratio varies wildly (10% to 80%)

Limitations: - Only works with local .claude directories (not remote sessions) - Session format is undocumented, might break in future - Analysis rules are heuristic-based

GitHub: https://github.com/yessGlory17/argus

Would love feedback, especially from other Claude Code users!

Comments URL: https://news.ycombinator.com/item?id=47228332

Points: 2

# Comments: 0

Article URL: https://twitter.com/thefp/status/2028587979900420303

Comments URL: https://news.ycombinator.com/item?id=47228314

Points: 2

# Comments: 0

I've been building for almost a year and sharing my journey on X, Reddit, and Hacker News. It seems like I still do not have the ability to break my bubble and reach more users. I really believe in the product I’m building, but acquiring users seems to be a harder task than building the product.

Any suggestions?

Comments URL: https://news.ycombinator.com/item?id=47228246

Points: 1

# Comments: 1

Over the past few weeks I’ve hit multiple AI/API outages — and not just one provider.

Model timeouts and HTTP 500/529 errors from Anthropic’s Claude

Repository access issues during a recent GitHub disruption

Database/API hiccups from Supabase

Even intermittent infra blips from major cloud vendors

Individually, none of these are shocking. Every system fails.

But collectively? It feels like the frequency and blast radius are increasing.

It raises some uncomfortable questions:

Are we over-optimizing for speed of development at the cost of resilience?

Are small teams unknowingly building on increasingly fragile stacks?

Is this just visibility bias — or are outages genuinely becoming more common?

Curious how others here are thinking about it.

Comments URL: https://news.ycombinator.com/item?id=47228213

Points: 18

# Comments: 5

I’ve been building a video-to-text web app and wanted to share it for feedback. The core flow is straightforward: upload files, start transcription, then track progress in a history page that refreshes automatically while jobs are running. Paid users can submit multiple files at once, and speaker diarization is supported for conversations and interviews.

Over the last few weeks I focused mostly on reliability. I changed the pipeline to extract audio first and then run transcription, which made long-file handling more stable. I also spent time improving failure handling so users see a clear message when a job fails, instead of raw model errors.

Pricing is intentionally simple right now: free users get 3 transcriptions per day, and there is one Unlimited plan at $20/month or $120/year.

I’d really appreciate feedback on the overall UX, whether the failure/retry behavior feels right, and whether the pricing is understandable for first-time users.

Comments URL: https://news.ycombinator.com/item?id=47228207

Points: 2

# Comments: 0

We are thrilled to share our preprint on PantheonOS, an evolvable, distributed multi-agent system for automatic genomics discovery.

Preprint: https://www.biorxiv.org/content/10.64898/2026.02.26.707870v1 Website(online platform free to everyone): https://pantheonos.stanford.edu/

PantheonOS unites LLM-powered agents, reinforcement learning, and agentic code evolution to push beyond routine analysis — evolving state-of-the-art algorithms to super-human performance.

Applications: Evolved batch correction (Harmony, Scanorama, BBKNN) and Reinforcement learning or RL agumented algorithms RL–augmented gene panel design Intelligent routing across 22+ virtual cell foundation models Autonomous discovery from newly generated 3D early mouse embryo data Integrated human fetal heart multi-omics with 3D whole-heart spatial data

Pantheon is highly extensible, although it is currently showcased with applications in genomics, the architecture is very general. The code has now been open-sourced, and we hope to build a new-generation AI data science ecosystem.

Comments URL: https://news.ycombinator.com/item?id=47228202

Points: 2

# Comments: 0

Article URL: https://co-r-e.github.io/dexcode-lp/

Comments URL: https://news.ycombinator.com/item?id=47228196

Points: 2

# Comments: 1

Article URL: https://www.alphaxiv.org/abs/2504.04372v3

Comments URL: https://news.ycombinator.com/item?id=47228188

Points: 1

# Comments: 0

I want to build an iOS app, although I already work as a software developer I'm afraid I can't afford one right now in my country. If anyone has an old Mac that you can gift me I would appreciate it!

Comments URL: https://news.ycombinator.com/item?id=47228179

Points: 3

# Comments: 1

Hi HN — I’m Prateek Rao. My cofounders and I built Cortexa, which we describe as a Bloomberg terminal for agentic memory.

A pattern I keep seeing: when agents misbehave, most teams iterate on prompts and then “fix” it by plugging in a memory layer (vector DB + RAG). That helps sometimes — but it doesn’t guarantee correctness. In practice it often introduces a new failure mode: the agent retrieves something dubious, writes it back to memory as if it’s truth, and that mistake becomes sticky. Over time you get memory pollution, circular hallucination loops, and debugging turns into log archaeology.

What Cortexa does:

1. Agent decision forensics (end-to-end “why”): trace outputs/actions back to the exact retrievals, memory writes, and tool calls that caused them.

2. Memory write governance: intercept and score memory writes (0–1), and optionally block/quarantine ungrounded entries before they poison future runs.

3. Memory hygiene + vector store noise control: automatically detect and remove near-duplicate / low-signal entries so retrieval stays high-quality and storage + inference costs don’t creep up.

Why this matters: Observability is the missing layer for agentic AI. Without it, autonomy is fragile: small errors silently compound, deployments become risky, and engineering cost goes up because failures aren’t reproducible or attributable.

Who this is for: 1. Teams shipping agentic workflows in production 2. Anyone fighting “unknown why” failures, memory pollution, or runaway context costs 3. Engineers who want auditability + faster debugging loops

Site: https://cortexa.ink/

Would love feedback from anyone running agents at scale: 1.What’s the most painful agent failure mode you’ve seen in production? 2.What signals would you want in an “agent terminal” (retrieval diffs, memory blame, tool-call traces, alerts, etc.)?

Comments URL: https://news.ycombinator.com/item?id=47228173

Points: 6

# Comments: 1