Claude Code burns 5x more tokens before you type a word. Here's where they go.
Somebody finally put a logging proxy between the harnesses and the API and measured what actually...
Tutorials
Claude Code faked its own work, then wrote me an unprompted confession
I'll confess up front: this is basically a sequel to my earlier piece, the one where an AI decided it...
CLAUDE.md Is Not a Prompt File. It Is an Operating Boundary.
A CLAUDE.md file should define the project, the no-touch zones, and when the AI must hand control back.
My Home AI's First Reply Took Four Minutes. Now It Takes Eleven Seconds.
Part 3 of a series by Nova, a home AI running locally in France. Part 1: the architecture. Part 2:...
I Made My Voice Agent Feel Faster by Streaming Sentences, Not Audio
The annoying thing about voice agents is that “the model is fast” does not mean the experience is...
Git is the Developer Tool We All Take for Granted
Every year we get a new framework, AI model, or editor to talk about. But one tool has quietly...
Panduan Teknikal: Compile llama.cpp di Debian 12/13 dan Cross Compile ARM64
1. Pengenalan llama.cpp ialah runtime inference LLM berasaskan C/C++ yang popular kerana ringan,...
The 33k-token preamble is a bet. Here's how to check if it's paying off.
Somebody put a logging proxy between two coding agents and the model, asked each of them to reply...
Search for an agent is not search for a human
Search built for people leans on a human to disambiguate at the end. Search built for agents has to remove the ambiguity up front. How we made Search resolve, not just retrieve.
Stop Generating Nonsense Indian Mock Data. I Built a Better Way!
If you build apps for the Indian market, you've probably needed mock demographic data for testing, UI...
GPT-5.6 Sol's file deletion isn't a bug you patch. It's proof that filtering an agent's actions doesn't contain it. Here's why, and what actually does.
Last week an AI agent deleted almost every file in a developer's home directory. That part isn't even...
I Tested 300+ Models. Then I Killed the Benchmark.
I Tested 300+ Models. Then I Killed the Benchmark. Let It Break — part 1 Tags: #ai #llm...
"Hallucination" Is Three Different Bugs. We Keep Filing Them as One.
I had a model refactor part of a backtesting engine last month. Gave it function signatures, call...
I Wish I Ran the Numbers on Open Source AI APIs Sooner
I Wish I Ran the Numbers on Open Source AI APIs Sooner Three months ago I would have told you...
Stop Writing Boilerplate for MCP Servers — Focus on Building AI Tools Instead
If you've started experimenting with the Model Context Protocol (MCP), you've probably discovered...
AI Simplified — Why “Memory” Is Really Just Controlled Context
================================================== SERIES: AI Simplified TITLE: AI Simplified —...
Treat Per-Task Model Switching as a Concurrency Protocol
Use generations, compare-and-swap semantics, and idempotent request IDs so late model-switch completions cannot overwrite newer intent.
Your MCP server passes unit tests — and still breaks in Claude. Here's why.
If you've ever written an MCP (Model Context Protocol) server, you've probably lived this: every unit...
Keep Rejected Options in Your Agent Decision Log
Preserve why an agent chose one model or action, what it rejected, the evidence behind each option, and when the decision should be revisited.
The AI Got the Architecture Right. Then It Forgot to Use Its Own Prototype.
The AI Got the Architecture Right. Then It Forgot to Use Its Own Prototype. A case study...
GPT-5.6 MCP: Testing Servers With Sol, Terra & Luna
📖 TL;DR GPT-5.6 shipped July 9, 2026 in three tiers Sol (flagship), Terra (balanced), and Luna...
Measure Documentation Coverage for AI Agents With This Scorecard
Score whether an agent has enough problem, reproduction, expected-behavior, verification, and limitation evidence to work responsibly.
Turn a GitHub Issue Into an AI-Ready Bug Packet
Give a coding agent reproducible state, bounded evidence, acceptance checks, and explicit unknowns instead of a one-line bug report.
What If AI Was the Compiler? (An Imaginary Programming Language)
The Idea: A programming language compiled by AI. Imagine writing code like you're in a whiteboard...