Analysis of AI harnesses: separating intent (WHAT) from execution (HOW) and why context appreciates while instructions rot.
Treating prompt changes like code deploys with eval gates to prevent silent failures in LLM-backed features.
How to automate AI prompt testing and compare models using promptfoo for reliable outputs.
Explains why externalizing judgment is the key skill for using LLMs effectively, beyond prompt engineering.
Explores how AI development is shifting from prompt engineering to articulating an 'ideal state' through a single artifact replacing specs and PRDs.
Explains why prompt libraries are insufficient for agentic AI and how to translate prompt intent into enforceable execution policies.
A system prompt designed to stop AI from mimicking human social patterns, promoting concise and technical responses.
How to use AI agents and worktrees to fix multiple GitHub issues on an old Rails app with zero manual effort.
Strategies to reduce GenAI and agent token costs while maintaining task quality through practical controls and telemetry.
Explains shifting from prompt engineering to intent engineering by focusing on outcomes rather than step-by-step instructions for AI.
A writer explains how AI-assisted writing provides access and voice, not slop or cheating, using personal essays as evidence.
Advice on keeping GitHub Copilot Agent Skills small and focused for better AI performance and predictable output.
Explores using an LLM to interview humans for context gathering, document creation, and review in complex tasks.
Structured-Prompt-Driven Development (SPDD) workflow for governable, reviewable, and reusable LLM-assisted code changes.
Learn two proven patterns for correctly using MCP servers with AI agents to avoid context bloat, higher costs, and poor performance.
Analyzes whether telling AI models 'You are an expert' improves responses, concluding clear context is more effective.
Transforming Anthropic's Claude system prompts into a git timeline for exploring prompt evolution via commit history.
Anthropic's Claude system prompts transformed into a git timeline for exploring prompt evolution via git log, diff, and blame.
Explores how naming AI chatbots creates distinct personalities, termed the 'Digital Ouija Effect', and its implications.
A critique of sharing AI chatbot screenshots, highlighting sycophancy, asymmetry of thought, and preference for human ideas over LLM outputs.
