Beyond Autocomplete: The Claude Code Boris Cherny Workflow Redefining 2026 Engineering
Discover the Claude Code Boris Cherny workflow. Learn how Anthropic's lead engineer uses 5 parallel agents, Opus 4.5, and automated verification loops to multiply productivity.
Silicon Valley isn't just listening—it's taking notes. Boris Cherny, the creator of Claude Code at Anthropic, just dropped a viral manifesto that changes everything about how we build software. What started as a terminal setup share has evolved into a paradigm shift: coding is no longer about syntax; it's about real-time strategy.
The Fleet Commander: Claude Code Boris Cherny Workflow
Cherny's secret isn't a faster keyboard—it's orchestration. He runs 5 parallel Claudes in his terminal at all times. Using iTerm2 system notifications, he manages multiple workstreams as a fleet commander. While one agent refactors legacy code, another runs a massive test suite, and a third handles documentation.
Choosing Smart Over Fast with Opus 4.5
In a world obsessed with latency, Cherny makes a counterintuitive choice: he exclusively uses Opus 4.5 with 'Thinking' enabled. Even though it's slower than Sonnet, its superior reasoning means less human steering. By paying the 'compute tax' upfront, he eliminates the 'correction tax' later, making the overall development cycle significantly faster.
Authors
Related Articles
Anthropic shipped Claude Sonnet 5, its mid-tier agent model, on June 30. Behind the 'Opus-class power, budget price' headline sits a tokenizer catch and a three-horse race that's only half real. Here's what developers should actually run the numbers on.
A critical vulnerability in Starlette—downloaded 325 million times per week—puts millions of AI agent servers at risk, exposing stored credentials for email, databases, and third-party services.
A small but growing group of developers has gone all-in on AI coding agents like Claude Code and OpenClaw. History suggests the rest of us won't be far behind.
Google is building AI agents that search the web proactively, without user prompting. That's not just a product update — it's a fundamental shift in who controls the information you receive.
Anthropic shipped Claude Sonnet 5, its mid-tier agent model, on June 30. Behind the 'Opus-class power, budget price' headline sits a tokenizer catch and a three-horse race that's only half real. Here's what developers should actually run the numbers on.
A critical vulnerability in Starlette—downloaded 325 million times per week—puts millions of AI agent servers at risk, exposing stored credentials for email, databases, and third-party services.
A small but growing group of developers has gone all-in on AI coding agents like Claude Code and OpenClaw. History suggests the rest of us won't be far behind.
Google is building AI agents that search the web proactively, without user prompting. That's not just a product update — it's a fundamental shift in who controls the information you receive.
Thoughts
Share your thoughts on this article
Sign in to join the conversation