Zero Trust for AI Agents: Stop Trusting Your Agent
Anthropic's Zero Trust for AI Agents, for engineers: ephemeral tokens, least agency, sandboxing, and which controls actually stop an automated attacker.
Library
All AI Coding Patterns
Technical articles: agents, architecture, tools and design decisions.
All articles
Git Worktrees and Claude Code: Parallel Agents Without Conflicts
How to use git worktrees to run multiple Claude Code instances in parallel on isolated checkouts: set up, integrate, and clean up without branches stepping on each other.
AI Examples by Industry: What It Does and Doesn't in Your Company
Realistic AI examples in retail, professional services, manufacturing and hospitality, each with its own risk. No hype, no inflated figures.
Git Best Practices in 2026: The Professional's Checklist
Atomic commits, clear messages, secrets kept out of the repo, and safe force pushes: the Git best-practices checklist for working with AI.
What Is an AI Guardrail and Why Do You Need One?
An AI guardrail is an external check that prevents an agent from deleting data, leaking information, or getting stuck in a loop. What it is and what types exist.
Team Git Workflow: Your First Day, Step by Step
The daily team Git cycle: updating main, creating a feature branch, small commits, opening a Pull Request, passing review, and merging.
AI at the center, not on top: when to redesign the process
Bolting AI on top of an old process pays off only halfway. When it makes sense to redesign the process with AI at the center, and when automating is enough.
What is a Skill?
A skill is a reusable capability you add to an AI assistant for a specific task. How it differs from a prompt, an agent, and an MCP.
Atomic Commits: Best Practices for a Clean History
What an atomic commit is, why it speeds up code review and makes revert and bisect easier, and how to split your work with git add -p step by step.
GitFlow vs GitHub Flow vs Trunk-Based: Which One to Choose
A senior comparison of the three branching strategies: tradeoffs, when to use each one, and how to migrate without breaking your team.
What Is a Claude Project and How Does It Work?
A Claude Project is a space in claude.ai and the app that stores instructions, files, and memory that each chat reuses without you having to re-explain anything.
New businesses AI makes viable that didn’t pay off before
AI doesn’t just make what you already do cheaper: it makes viable businesses and services that never covered their cost. Where the real opportunity is, no hype.