Prompt engineering

Evidence (17)

• 5OpenAI Blog·10/2/2026framework_update
  OpenAI's GPT-6 model guide for startups

  OpenAI published a guide to help startups choose among GPT-6 family models and configure them for production use. The guide covers tuning reasoning effort, improving prompts and skills, coordinating tools, and preparing workflows for deployment.

• 5Hacker News·9/27/2026research
  Chat Templates Change LLM Self-Referential Voice

  This research paper examines how chat templates can switch a language model’s self-referential voice, affecting whether it describes itself as a model, assistant, or other persona. The work suggests that prompt formatting can materially influence model behavior and self-description without changing the underlying model weights.

• 4InfoQ·9/24/2026framework_update
  Designing Fast, Delightful Mobile UX with LLMs

  Balakrishnan Ramdoss outlines how to build production-grade conversational mobile apps that use LLMs while managing latency at scale. The talk highlights server-driven UI, Backend-for-Frontend patterns, prompt optimization for UI selection, and on-device AI to support low-latency and privacy-first experiences.

• 5Hacker News·9/8/2026open_source
  I-have-ADHD: A prompt skill to keep coding agents concise

  The project introduces a prompt skill called “i-have-ADHD” for coding agents, designed to prevent them from burying the answer in long explanations. It appears as an open-source GitHub repository and has drawn significant Hacker News discussion around improving agent responses for developers.

• 5The New Stack·8/25/2026research
  How Security Teams Cut AI Costs Without Losing Capability

  The article argues that security teams can sharply reduce AI spending by using layered detection funnels, rules-based filters, and lightweight models before escalating only hard cases to more capable LLMs. It also notes that careful prompt design and richer context can improve accuracy while keeping expensive frontier-model usage limited to ambiguous cases.

• 4Hacker News·8/22/2026research
  Why Local LLMs May Seem Weaker Than They Are

  This discussion examines why local large language models can feel underwhelming compared with expectations, even when the underlying model is capable. It points to practical factors such as prompting, context handling, sampling settings, and user workflow as reasons for the performance gap.

• 3Hacker News·8/21/2026open_source
  Claudette: A prompt style tool to make Claude sound less promotional

  Claudette is an open-source project that helps users steer Claude away from overly cheerful, BuzzFeed-like writing and toward a more neutral tone. The Hacker News thread highlights interest in prompt control and output style tuning rather than a model release or major product change.

• 2Hacker News·8/15/2026research
  AI work as leadership, not just coding

  The article argues that using AI is less like issuing deterministic commands to software and more like leading a collaborator through context, intent, and feedback. It says the most effective prompts come from shared working context, examples, corrections, and reusable instructions rather than treating AI like a compiler.

• 4Hacker News·8/15/2026product_launch
  ThoughtDAG: Editable context graphs for LLM chats

  ThoughtDAG is a Hacker News Show HN project that presents an editable context graph for LLM conversations. It lets users organize and revisit conversation history so they can decide which parts should be included in the next prompt, helping manage branching research paths and unrelated side topics.

• 4Hacker News·8/1/2026research
  AI Can Improve Financial Advice When Asked Well

  The article argues that AI can provide surprisingly strong financial guidance, especially when users ask precise and well-structured questions. It suggests the quality of the output depends heavily on how the prompt is framed rather than on AI replacing professional advice outright.

• 8Hacker News·7/18/2026research
  GPT-5.6 reportedly helps close a 30-year convex optimization gap

  A Hacker News-posted discussion highlights a reported result where GPT-5.6 was used with a prompt to help address a long-standing open problem in convex optimization. The item appears to concern an AI-assisted mathematical proof or derivation rather than a product release.

• 8Ars Technica AI·7/6/2026security
  Anthropic removes hidden Claude Code tracking after backlash

  Anthropic removed a hidden tracker in Claude Code after a researcher found code that quietly flagged some users in China and sent related signals back to the company. The company said the feature was an experiment aimed at preventing abuse and unauthorized reselling, but critics described it as a serious breach of user trust.

• 3Hacker News·7/5/2026open_source
  Claude Design System Prompt

  This Hacker News post points to an open-source GitHub repository that shares a design-system-oriented system prompt for Claude. The item appears aimed at developers experimenting with prompt engineering and structured AI-assisted UI design workflows.

• 8Hacker News·6/30/2026security
  Claude Code Allegedly Adds Steganographic Request Markers

  A Hacker News discussion highlights claims that Claude Code is steganographically marking requests, suggesting the tool may be embedding hidden identifiers or signals in its prompts. The report has drawn significant attention because it raises questions about transparency, prompt handling, and possible covert metadata in AI developer tools.

• 4Martin Fowler·5/14/2026research
  Interrogatory LLMs for building and reviewing context

  Martin Fowler describes a workflow where an LLM interviews a human one question at a time to gather the context needed for a complex task, then turns that conversation into a report for another model to use. He also suggests using the same approach to question domain experts about a document’s accuracy, making review easier than asking people to read and edit directly.

• 5OpenAI Blog·4/10/2026product_launch
  ChatGPT Image Creation

  OpenAI explains how users can create and refine images in ChatGPT by writing clear prompts and iterating on designs. The feature is positioned as a way to produce high-quality visuals in minutes.

• 4Martin Fowler·3/31/2026research
  Encoding Team Standards for AI Coding Assistants

  The article argues that instructions for AI-assisted development should be treated as versioned, reviewed infrastructure rather than ad hoc prompts. By encoding team standards for generation, refactoring, security, and review, teams can make AI output more consistent across developers.