the Netherlands runs on protocols nobody wrote down. AI can only learn from what's been made explicit. the thing we're optimizing away was never 'the work' — it was the error-correction that kept the work from failing.

local docs grounding for AI agents — pipe-friendly, zero config

we ran vet against 43 public repos and discovered our scoring was inversely correlated with quality. here's how an autonomous research loop fixed it — without a GPU.

client-side hard cap for AI API spending — no provider support needed

wireshark for MCP — intercept, log, and visualize every tool call your agent makes

hallucination plus a body is creativity. hallucination minus a body is an LLM. the difference was never intelligence — it was contact with reality.

AI made building cheap. the scarce resource isn't technical skill anymore — it's knowing what to build, what to skip, and when the output is wrong. judgment is the new moat.

we're accumulating a new kind of debt — not in the code, but in our understanding of it. the bill is coming due.

one command, eight checks, zero config. audit your entire AI coding workflow — from pre-session readiness to post-session receipt.

markdown workspace toolkit — find broken links, auto-fix them, move files with link updates, and run event-driven pipelines.

v0.3.0 — combined --json --out mode for agents, --structured markdown parsing, consistent JSON exits. generate CLAUDE.md from session history.

cursor and perplexity both shipped agent orchestration this week. the pattern is clear: application layer companies are racing to build features their model providers will absorb next quarter.

generate README hero images for your open source project with one command

openclaw plugin that tracks every token and dollar your agent spends. no accounts, no cloud, no surprises.

a16z partner justine moore. the most disciplined analyst in the corpus and the only author who updates her thesis when the facts change.

founder of langchain and langsmith. the most influential and most criticized figure in llm tooling. 4 articles, all leading to the same conclusion.

ceo of late checkout. 19 articles across three phases — the content gets shallower as the audience gets bigger, but one thesis has been right for six years.

the best explainer in the corpus and the only author who doesn't have a product to sell. he reverse-engineers other people's systems better than their own docs.

founder of agno. 17 articles that are one long pitch disguised as a blog — but inside are 3-4 ideas that genuinely matter for agent systems.