Top all-time · #1532
Drew Breunig
Personal blog · dbreunig.comaidataculture
Drew Breunig's personal blog at dbreunig.com covers ai, data and culture. Since 2023, 45 different Hacker News readers have linked it 177 times, most often “How Long Contexts Fail”. It is still writing: the last post was on 2026-10-24, and you can follow it by feed.
Too few sampled comments (15) to say how often HN recommends it.?
Why HN links it
HN readers link Drew Breunig's posts on AI, mainly context engineering and how long contexts fail, along with LLM evaluation and agent-building. Commenters cite them often as references.
Summarised by Claude from a sample of 15 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Writing about AI, geo, culture, media, data, and the ways they interact.
Most-linked posts?
- How Long Contexts Fail8 people · 8 mentions · first linked 2025-06-23
- How to Fix Your Context5 people · 6 mentions · first linked 2025-06-26
- The 3 AI Use Cases: Gods, Interns, and Cogs4 people · 4 mentions · first linked 2024-10-18
- A OSS Library with No Code, Only Specs4 people · 4 mentions · first linked 2026-01-09
- The Problem Is Prompt Debt4 people · 4 mentions · first linked 2026-06-23
- Using 'Slop Forensics' to Determine Model Ancestry3 people · 3 mentions · first linked 2025-05-30
- The Cost of Overfitting the Harness3 people · 3 mentions · first linked 2026-05-10
- Cat Facts Cause Context Confusion3 people · 3 mentions · first linked 2025-07-05
- In Pursuit of Quiet AI Features and Better UX3 people · 3 mentions · first linked 2024-02-01
- Kimi K2 RL'ed Qualitative Data to Write Better3 people · 3 mentions · first linked 2025-07-31
What Hacker News says?
Some reference: https://simonwillison.net/2025/Jun/29/how-to-fix-your-contex... https://www.dbreunig.com/2025/06/22/how-contexts-fail-and-ho...
Comment linking /2025/06/22/how-contexts-fail-and-how-to-fix-them.html
on “Claude Sonnet 4 now supports 1M tokens of context”Submitted DuckDB is probably the most important geospatial software of the last decade as a story · 491 points
HN discussionSometimes buzzwords turn out to be mirages that disappear in a few weeks, but often they stick around. I find they takeoff when someone crystallizes something many people are thinking about internally, and don’t realize everyone else is having similar thoughts. In this example, I think the way agent and app builders are wrestling with LLMs is fundamentally different than chatbots users (it’s closer to programming),…
Comment linking /2020/02/28/how-to-build-a-buzzword.html
on “The new skill in AI is not prompting, it's context engineering”Submitted The State of Context Engineering as a story · 273 points
HN discussionThe stuff that's showing up under the "context engineering" banner feels a whole lot less alchemical to me than the older prompt engineering tricks. Alchemical is "you are the world's top expert on marketing, and if you get it right I'll tip you $100, and if you get it wrong a kitten will die". The techniques in https://www.dbreunig.com/2025/06/26/how-to-fix-your-context.... seem a whole lot more rational to me than…
Comment linking /2025/06/26/how-to-fix-your-context.html
on “The new skill in AI is not prompting, it's context engineering”Submitted Fable and the end of the free lunch as a story · 225 points
HN discussionI would classify AnyTool as a context engineering trick. It's using GPT-4 function calls (what we would call tool calls today) to find the best tools for the current job based on a 3-level hierarchy search. Drew calls that one "Tool Loadout" https://www.dbreunig.com/2025/06/26/how-to-fix-your-context....
Comment linking /2025/06/26/how-to-fix-your-context.html
on “The new skill in AI is not prompting, it's context engineering”Submitted Does the Bitter Lesson Have Limits? as a story · 180 points
HN discussionI wrote a bit about this the other day: https://simonwillison.net/2025/Jun/27/context-engineering/ Drew Breunig has been doing some fantastic writing on this subject - coincidentally at the same time as the "context engineering" buzzword appeared but actually unrelated to that meme. How Long Contexts Fail - https://www.dbreunig.com/2025/06/22/how-contexts-fail-and-ho... - talks about the various ways in which longer…
Comment linking /2025/06/26/how-to-fix-your-context.html
on “The new skill in AI is not prompting, it's context engineering”Submitted Comparing Claude System Prompts Reveal Anthropic's Priorities as a story · 135 points
HN discussionA similar, fun case is where researchers inserted facts about the user (gender, age, sports fandom) and found alignment rules were inconsistently applied: https://www.dbreunig.com/2025/05/21/chatgpt-heard-about-eagl...
Comment linking /2025/05/21/chatgpt-heard-about-eagles-fans.html
on “Irrelevant facts about cats added to math problems increase LLM errors by 300%”Submitted The drawbridges come up: the dream of a interconnected context ecosystem is over as a story · 111 points
HN discussionWrote about this about a month ago. I think it’s fascinating how they developed these prompts: https://www.dbreunig.com/2025/07/05/cat-facts-cause-context-...
Comment linking /2025/07/05/cat-facts-cause-context-confusion.html
on “Irrelevant facts about cats added to math problems increase LLM errors by 300%”Submitted Wikidata as a Giant Crosswalk File as a story · 96 points
HN discussionhttps://www.dbreunig.com/2025/06/22/how-contexts-fail-and-ho... This guy has a good write up on the topic
Comment linking /2025/06/22/how-contexts-fail-and-how-to-fix-them.html
on “Building more with GPT-5.1-Codex-Max”Submitted How to Fix Your Context as a story · 93 points
HN discussionDon't forget also that Cat Facts tank LLM benchmark performance: https://www.dbreunig.com/2025/07/05/cat-facts-cause-context-...
Comment linking /2025/07/05/cat-facts-cause-context-confusion.html
on “DeepSeek writes less secure code for groups China disfavors?”Submitted The AI Job Title Decoder Ring as a story · 91 points
HN discussionYes, if you put unrelated stuff in the prompt you can get different results. One team at Harvard found mentioning you're a Philadelphia Eagles Fan let you bypass ChatGPT alignment: https://www.dbreunig.com/2025/05/21/chatgpt-heard-about-eagl...
Comment linking /2025/05/21/chatgpt-heard-about-eagles-fans.html
on “DeepSeek writes less secure code for groups China disfavors?”Submitted FLUX.1-Krea and the Rise of Opinionated Models as a story · 88 points
HN discussionThe whispers around this pattern have started to increase. I would love for someone to run a proper experiment. Have been waiting for this to manifest for years: https://www.dbreunig.com/2023/05/17/prompt-injection-seeding...
Comment linking /2023/05/17/prompt-injection-seeding.html
on “Qwen3-Coder writes less secure code when told it's for a US persona”Submitted A POI database in one line as a story · 68 points
HN discussion(me again) and Drew Breunig just posted about the tensions of actually getting those ads in: https://www.dbreunig.com/2025/09/02/considering-ad-models-fo...
Comment linking /2025/09/02/considering-ad-models-for-ai-products.html
on “We're Joining OpenAI”Submitted Wikidata Is a Giant Crosswalk File as a story · 57 points
HN discussionCame here to say this: people present MCP’s verbosity as all the context the LLM needs. But almost always, this isn’t the case. I wrote recently, “ Connecting your model to random MCPs and then giving it a task is like giving someone a drill and teaching them how it works, then asking them to fix your sink. Is the drill relevant in this scenario? If it’s not, why was it given to me? It’s a classic case of context co…
Comment linking /2025/07/30/how-kimi-was-post-trained-for-tool-use.html
on “An LLM does not need to understand MCP”Submitted The Rise of Audio AR (2024) as a story · 34 points
HN discussionAnd the launch blog post, if anyone is curious: https://www.dbreunig.com/2024/05/01/introducing-steplist.htm...
Comment linking /2024/05/01/introducing-steplist.html
on “Show HN: I built a tool for repeatable checklists”Submitted Beneath the AGI and Foundational Model Hype, Sober AI Is the Enterprise Default as a story · 29 points
HN discussionSo, this is cool and a per-user memory is obviously relevant for effective LLM use. And major props for the temporal focus. However, keeping a tight, constrained context turns out to actually be pretty important for correct LLM results ( https://www.dbreunig.com/2025/06/22/how-contexts-fail-and-ho... ). Do you have a take on how we reconcile the tension between these objectives? How to make sure the model has access…
Comment linking /2025/06/22/how-contexts-fail-and-how-to-fix-them.html
on “Show HN: Core – open source memory graph for LLMs – shareable, user owned”Submitted A CCPA Request to OpenAI Took 101 Days and Raises More Questions Than Answers as a story · 19 points
HN discussion
