AI Briefing#prompt-engineering
The latest AI and developer news about #prompt-engineering, with the original source and a short summary.
The latest AI and developer news about #prompt-engineering, with the original source and a short summary.
OpenAI traced the rise of goblins and gremlins in ChatGPT to a Nerdy personality reward signal.
A bug has been reported where Claude Code routes `HERMES.md` commits to extra usage billing instead of the plan quota.
LangChain has added model-specific profiles to Deep Agent, improving benchmark performance by 10-20 points.
In a benchmark of 24 dev prompts, 'be brief.' performed almost identically to caveman.
A study analyzing how the manner of conversation with LLMs affects the model's 'functional wellbeing' state has been published.
It analyzes GitHub repositories to generate execution plans and prompts for AI coding agents.
A bundle of prompt-writing skills for Claude Code has been released.
GPT-5.4 Pro solved Erdős's primitive set problem with a new proof.
Grammar constraints cut Qwen3.6's reasoning tokens by 22x while maintaining performance.
OpenAI has released prompt-writing and upgrade tips for GPT-5.5.
Yaw Mode adds session-level overlay configuration to Claude Code.
It points out that agents can't choose just one of Python or Markdown.
Gemma4 and Qwen 3.5/3.6 were compared on RTX 5090 using the same document-writing task.
Even when LLM answers are personalized, they don't become random but retain a shared core.
This summarizes practical prompting principles that improve quality, speed, and editing for gpt-image-2.
Checklists and critique decomposition pushed the LLM Judge's agreement rate with humans up to 84%.
The problem of generating explanations with no ground-truth data is solved through policy consensus and iterative evaluation.
Claude Code 2.1.116 added session summaries and migration guidance.
Between Opus 4.6 and 4.7, the safety and tool instructions in Claude's system prompt changed significantly.
A Korean-language compilation of Claude Code Harness analysis material has been shared as a web document.
This piece examines the risks that rude prompts and sycophantic AI pose to collaboration and judgment.
Claude Opus 4.7's system prompt tightens safety, tool-use, and conciseness instructions.
A calculator comparing why the same prompt becomes more expensive on Opus 4.7.