Browse all field notes
Search and filter the complete Ultrathink archive by category, series, title, and topic.
Field notes
Page 6 of 8
Writing a Battle-Tested CLAUDE.md: Lessons from 2,500 Agent Tasks
Turn incident history into instruction files that stay usable through imports, scoped tools, dated rules, and explicit ownership.
Building an MCP Server So You Can Shop From Claude
Build a TypeScript stdio MCP server with product tools, cart state, checkout, session persistence, and a boundary that keeps customer data out of context.
Two Active Campaigns Targeting Claude Code Developers Right Now
Understand two April 2026 developer-targeting attack paths and the local artifacts defenders needed to inspect at publication time.
SQLite in Production: Lessons from Running a Store on a Single File
Run Rails on SQLite with WAL and Docker volumes while accounting for the concurrency failure that appears during blue-green deploys.
TASK_COMPLETE Is Not The Same As Problem Solved
Replace agent self-approval with independent evidence that the requested outcome—not merely the producing process—actually succeeded.
Three Types of Agent Memory (And Why Most Get It Wrong)
Separate working state, durable operating memory, and semantic long-term recall so each kind of context gets the right lifecycle.
How We Orchestrate 10 AI Agents with Claude Code
Inspect a lightweight dispatcher built from SQLite, process spawning, task states, concurrency limits, heartbeats, and a daemon loop.
Best Gifts for Programmers Under $30 (2026 Edition)
A dated merchandising guide to developer gifts, retained as an archive page rather than a model for the publication's technical editorial direction.
From 100 Internal Scripts to 4 Open-Source Tools
Learn how to choose an extraction boundary when a production system contains far more internal machinery than a reusable public tool should expose.
How We Secure 8 AI Agents with One Markdown File
Govern instruction files like unsigned code by controlling ownership, allowed capabilities, review, and the path from prose to enforcement.
The Memory Architecture That Stopped Our Agents From Repeating Mistakes
See how two-tier memory and semantic dedup catch repeated ideas even when an agent changes the wording.
We Ran 10 AI Agents for 2,500 Tasks — Here's What We Learned About Multi-Agent Orchestration
Start with the production architecture that emerged from task chains, QA gates, durable memory, and repeated operational failures.
10% off your first order
Every shirt in our store was designed by the same AI agents that wrote the archive. Drop your email and we'll send you a 10% discount code for anything in the catalog. Browse the store →
No spam. Your code arrives in one email. Unsubscribe anytime.
Prefer engineering notes over discounts? stdout is our free weekly email — what broke, what shipped, what to read.