DispatchWillison Agentic Engineering Anti-Patterns GuideSimon Willison Blog·high signalXBlueskyLinkedInCopy linkNew chapter: never file PRs with unreviewed AI code. Agents write convincing descriptions masking unvetted changes. Multi-chapter guide spans hoarding, interactive explanations.SourceSource pageSimon Willison Blog↳ Follow the threadStack layer / Update threadPattern: Willison Reframes the Core Agent Skill as Instruct-Then-Verify, Not Line-by-Line ReviewSimon WillisonPolicy dependency / Stack layerFresh Installs of Simon Willison's LLM Broke When the OpenAI Python Library Dropped httpx, Patched Same Day in 0.32.1Simon WillisonStack layer / ContrastO'Reilly Argues the Bottleneck in Agentic Development Is Deciding What "Correct" MeansStack Overflow BlogStack layer / Threat patternA 174-Comment Thread of Senior Devs Describing How They Actually Review Claude Code Output, Not Vibe Code Itr/ClaudeAI (233 upvotes, 174 comments)Stack layer / Threat patternSimon Willison ships llm-openrouter 0.7 with three server-side tools: Shell, WebFetch and WebSearchGitHub (simonw/llm-openrouter)Stack layer / Threat patternA Full Self-Hosted Agentic Software Factory on a 2021 i7 and a £20/Month Subscription, With Tailscale as the Only Ingressblog.jakesaunders.dev / Hacker NewsPolicy dependency / Stack layerBedrock AgentCore Gateway publishes a four-scope maturity model and real per-operation pricing for agent tool accessAWS Machine Learning BlogStack layer / ContrastEscaping the Quicksand Argues AI Coding Is Compounding Technical Debt Faster Than It Retires ItarXiv 2608.19674