frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

What I changed in how I use Claude Code after Anthropic's postmortem

4•cinooo•1h ago
After watching Anthropic's recent postmortem (anthropic.com/engineering/april-23-postmortem), I've been thinking about the way I approach Claude Code differently. They lowered default reasoning effort to fix latency, called it the wrong tradeoff, and reverted under public scrutiny. The reverts are good but they don't change the underlying equation I'd been ignoring.

The fact is we potentially have a team of engineers at our disposal now. Token cost is a real cost, but I couldn't hire my own engineer for any of the personal work I'm doing. Apply that lens to token usage and the approach shifts. It stops being about the cost ceiling and becomes a cost/output/quality view, the same way I'd think about hiring decisions on a real team.

Four places I now think about: model, configuration, prompting, agents.

On model. Opus is still the strongest for critical decisions and architectural reasoning. Sonnet is usually good enough for coding and simple repetitive tasks. I use the right model for the work. If I cheap out, I can't expect quality.

On configuration. /effort runs from low to max. Opus 4.7's default is xhigh. I set the level to fit the work. A quick edit doesn't need max, an architectural decision does. The cheapest move and the one I'd been skipping.

On prompting: three patterns I find most effective.

1. "Ask questions if unsure." Without this I'm not giving the model an out, which closes off solutions even when there's no clean answer and tradeoffs need to be surfaced.

2. "Time and cost are not factors here. Prefer robust, sustainable, scalable solutions, do not leave tech debt." This inverts the implicit optimisation pressure for the duration of the task.

3. "Reflect on this session and encode via claude.md or skills what you learned, so the next iteration doesn't repeat the same mistakes." Worth capturing as a skill and iterating for myself. Without this every session starts from zero, repeating mistakes I've already corrected.

On agents. Without going into detail as this is a whole post in itself, the pattern that works for me is using agents to separate concerns. One agent does spec review against the code (code is source of truth). A separate agent does code review after implementation.

Engineering and product teams have always balanced speed to market with cost and quality. AI is no different. The difference is which levers I choose. Spend the budget on effort deliberately, and the work comes back at the level I want.

Humanoid Robots Are Still a Body Problem with Jerry Pratt [video]

https://www.youtube.com/watch?v=fsPDZjWMKMs
1•thunderbong•1m ago•0 comments

Meta stock drops on quarterly results 'internet disruptions' user numbers down

https://www.cnbc.com/2026/04/29/meta-q1-earnings-report-2026.html
1•1vuio0pswjnm7•5m ago•0 comments

In Backlash Against Tech in Schools, Parents Are Winning Rollbacks

https://www.nytimes.com/2026/04/29/technology/parents-school-tech-backlash.html
1•1vuio0pswjnm7•8m ago•0 comments

Tourist dies after being bitten at snake show while on vacation in Egypt

https://www.cnn.com/2026/04/28/africa/tourist-dies-snake-bite-egypt-scli-intl
1•fortran77•9m ago•0 comments

OpenAI, Sam Altman Hit with Slate of Lawsuits over Mass Shooting Canadian School

https://www.law.com/therecorder/2026/04/29/openai-sam-altman-hit-with-slate-of-lawsuits-over-mass...
1•1vuio0pswjnm7•10m ago•0 comments

Infrastructure Spend in the AI Era

https://redmonk.com/sogrady/2026/04/29/infrastructure-spend-in-the-ai-era/
1•mooreds•10m ago•0 comments

The Harmful Consequences of the Robustness Principle (2018)

https://datatracker.ietf.org/doc/html/draft-thomson-postel-was-wrong-03
1•Tomte•11m ago•0 comments

AMS – Detect unsafe LLMs in 30 seconds via activation analysis

https://github.com/GoogleCloudPlatform/activation-model-scanner
1•gmessenger•14m ago•0 comments

Fanspeak (1999)

https://www.fantasymaps.com/stuff/fanspeak.html
1•ipnon•14m ago•0 comments

Operation Sundevil (1990)

https://en.wikipedia.org/wiki/Operation_Sundevil
1•rickcarlino•20m ago•0 comments

Stablecoin Reserve Race: Brazil Pix, Morgan Stanley, OCC Charters

https://stablecoinbrief.substack.com/p/stablecoin-reserve-race-brazil-pix
1•knivef•26m ago•0 comments

Building agents that reach production systems with MCP

https://claude.com/blog/building-agents-that-reach-production-systems-with-mcp
1•gmays•32m ago•0 comments

Biology is a Burrito: A text- and visual-based journey through a living cell

https://burrito.bio/essays/biology-is-a-burrito
18•the-mitr•46m ago•2 comments

Can LLMs create lasting flashcards from readers' highlights?

https://memory-machines.com/report/
1•jryio•46m ago•0 comments

The Arpanet Reconstruction Project

https://obsolescence.dev/arpanet_home.html
2•ecliptik•47m ago•0 comments

Where the Goblins Came From

https://openai.com/index/where-the-goblins-came-from/
107•ilreb•49m ago•33 comments

Demonstrating the idea of gamma camera imaging [video]

https://www.youtube.com/watch?v=PyGlHtvihXA
5•num42•51m ago•0 comments

Ernie 5.1 Preview

https://ernie.baidu.com/
3•qainsights•52m ago•0 comments

Strait of Hormuz Daily Intelligence

https://insights.windward.ai/
3•jnord•54m ago•0 comments

LFM2-24B-A2B: Scaling Up the LFM2 Architecture

https://www.liquid.ai/blog/lfm2-24b-a2b
2•nateb2022•55m ago•0 comments

Finetuning Activates Verbatim Recall of Copyrighted Books in LLMs

https://github.com/cauchy221/Alignment-Whack-a-Mole-Code
16•reconnecting•58m ago•1 comments

Functional Programmers need to take a look at Zig

https://pure-systems.org/posts/2026-04-29-functional-programmers-need-to-take-a-look-at-zig.html
15•xngbuilds•59m ago•1 comments

Transponders to be installed on New York area airport ground vehicles

https://apnews.com/article/laguardia-plane-crash-air-canada-transponders-18578e94f2f7ec67b5ed45e9...
2•geox•1h ago•0 comments

GitHub Is Sinking

https://dbushell.com/2026/04/29/github-is-sinking/
5•xngbuilds•1h ago•1 comments

Opus 4.7's New Tokenizer: What It Costs

https://openrouter.ai/announcements/opus-47-tokenizer-analysis
4•vinhnx•1h ago•0 comments

Scroll-Driven Animations

https://www.joshwcomeau.com/animation/scroll-driven-animations/
2•vinhnx•1h ago•0 comments

AI Groupchats app just launched

https://techcrunch.com/2026/04/29/meet-shapes-the-app-bringing-humans-and-ai-into-the-same-group-...
2•nooriee•1h ago•1 comments

Show HN: Lssh – Terminal-native remote access suite for SSH and cloud targets

https://github.com/blacknon/lssh
2•blacknon•1h ago•1 comments

What is something you started and then wish you started earlier?

https://old.reddit.com/r/AskReddit/comments/1szjkl5/comment/oj2a8b7/
3•eeko_systems•1h ago•0 comments

I Choose Email over Messaging

https://www.spinellis.gr/blog/20250926/
3•fagnerbrack•1h ago•0 comments