AI Intelligence Archive

Many RAG failures that look like model problems are ingestion problems. If a parser flattens a PDF...

Deterministic contract tests for structured LLM output on the JVM, and what 120 real documents taught...

A model writes the change. A model reviews the change and says it looks good. The pull request now...

Why an agent's freshly written helper often duplicates an existing utility, and how silent forks drift into real bugs.

git commit, npx, apt, ssh, psql, terraform apply, gh auth login: each can stall an AI coding agent on a prompt. The flag or env var that stops each one.

Claude Looked Blocked. My Policy Had Never Fired. From the agentpolicy v0.1.0 release: a...

10 Prompts That Turn ChatGPT Into a Senior Data Analyst I once spent an entire weekend...

A 6B design model peaked at 64.4 GiB on an A100 80GB. Its transparent PNGs depend on a line break, so verify the alpha channel yourself.

When you ask Claude Code to "make a 60-second horror reel", it will happily start writing components....

We replayed 45 findings from a careful human code review against automated rules. They caught 9. Here is what that taught us about which problems tools can own, and the toolkit we built around it.

Developers discover that Claude Code ignores custom instructions because they set them in the wrong layer. Learn the three ways to shape Claude Code behavior, what the system prompt actually controls, and which parts remain locked no matte…

Getting a language model to answer WhatsApp messages takes an afternoon. Getting it to stop answering...

The problem nobody talks about AI agents are getting persistent memory. mem0, Zep, Letta,...

A tampered Claude Code extension sat in Cursor's marketplace for about 11 hours. Here is the delivery chain that made it look like a normal update.

Originally published by Alicia on 2 November 2026 at Digital Garden. Original measurements and...

On Opus 5.5 a warm 200K-token Claude session reads its history for $0.04. Let the cache expire and the next line pays $1.00 to write it back. Do the math.

Disclosure: I work on FORMAT, a tool that turns interior renders into short videos. The keyword data...

If an AI writes your code, your tests are the only thing standing between "it compiles" and "it's...

Train a toy byte-pair encoder in TypeScript and print every merge. Then ask o200k_base why pretty JSON and long keys cost you more tokens.

"Will it fit on my card?" comes up constantly in local-LLM threads, and the usual answer is a rule of...

This is a submission for the MLH x DEV Writing Challenge Finding a food pantry is one step. Knowing...

Last month I finished one of the largest side project of my career — an open-source framework, docs,...

Agents can retype a module in an afternoon. Joel Spolsky's rule against rewrites was about the bug fixes buried in old code. Pin those first.

Nobody changed the model. Nobody touched the prompt. Accuracy dropped anyway. This is the most...

A screenshot can show what you built. A screen recording can show how it works. But making a short,...

Both make a model answer faster. Only one changes what the model is. A cost-and-quality comparison for teams shipping latency-sensitive inference.

Long context is sold as a feature and paid for as a memory bill. A cost-first look at why KV cache compression beats simply adding windows.

I coded by hand for about three years. When a feature was done, I could close the file and still...

Twenty agent turns, zero commits, red tests. A Claude Code Stop hook snapshots each turn to refs/agent-turns/, then git bisect finds the turn that broke it.

Someone stole a dead man's portrait — and in a village where being remembered is what keeps you real,...

Apollo's chief economist argued in late October that AI has delivered "zero financial payoff" to...

A zinc sunscreen's label tells you how much zinc oxide is in it. It doesn't tell you how white it's...

ADD FOREIGN KEY takes a write-blocking lock on both tables while it scans. Reproduce it in Docker, then split it into NOT VALID and VALIDATE.

TL;DR: In 2013 I wrote a university diploma on elliptic-curve cryptography, and it went into a...

Disclosure: I build PartsPlan, the tool mentioned at the end. The cases below are real answers I...

Wiring Your AI Agent into a "Mycelium Network": A Zero-Dependency MCP Server in...

A PR shows the branch, lazygit shows hunks, and neither knows which agent turn wrote a line. A fair look at three places to review agent code.

Every orchestration framework I used managed calls. None of them managed the organization. So I...

Seven nights, 8,654 bot knocks, 247 human visits. AI crawlers vs search crawlers, measured: Perplexity beat OpenAI 9-to-1 and 65% knocked unnamed.

If you've built an MCP server or an API that agents call, you've run into this problem. Your agent...