Claude Code
Agentic Coding with ClaudeOfficial best practices, creator interviews, and real-world workflows for Anthropic's agentic coding CLI, Claude Code.
About this vault
A curated set of articles on Claude Code: Anthropic's own engineering guidance on best practices, large-codebase adoption, hooks, and subagents; interviews and podcast conversations with creator Boris Cherny and PM Cat Wu on how the tool was built and how it's used inside Anthropic; and hands-on workflow writeups from practitioners covering research-plan-implement loops, parallel agents, git worktrees, and MCP tooling.
12 articles
Best practices for Claude Code
Anthropic's official engineering guide to Claude Code, covering the core constraint that context window fill-up degrades performance, and practical patterns like giving Claude a verifiable check, exploring/planning before coding, and writing precise prompts with rich context.
How Claude Code works in large codebases: Best practices and where to start
An Anthropic blog post explaining why Claude Code's agentic, index-free search (file traversal, grep, references) holds up better than RAG-based retrieval in large, fast-moving codebases, and detailing the five-layer 'harness' — CLAUDE.md, hooks, skills, plugins, and MCP servers, plus LSP integrations and subagents — that determines how well Claude Code performs at scale.
Hooks reference - Claude Code Docs
The official Claude Code reference for hooks — user-defined shell commands, HTTP endpoints, or LLM prompts that fire automatically at specific points in a session's lifecycle. Documents the full event list (SessionStart, PreToolUse, PostToolUse, Stop, SubagentStart/Stop, PreCompact, etc.), JSON input/output schemas, exit-code behavior, and async/HTTP/agent-based hook types.
Create custom subagents - Claude Code Docs
Official Claude Code documentation on subagents — specialized AI assistants that run in their own context window with their own system prompt, tool access, and permissions. Covers the built-in Explore/Plan/general-purpose agents, how to write and scope custom subagent files (.claude/agents/ vs ~/.claude/agents/), and common patterns like isolating high-volume operations and running parallel research.
Agentic coding and persistent returns to expertise: How Claude Code is used in practice
Anthropic economic research analyzing ~400,000 privacy-preserving Claude Code sessions from October 2025 to April 2026, finding a stable division of labor (people decide what to build, Claude decides how), rising task value and success rates over time, and that domain expertise — not coding skill — is the strongest predictor of a session succeeding.
Claude Code: Anthropic's Agent in Your Terminal
Latent Space podcast episode and transcript with Cat Wu (PM) and Boris Cherny (creator) of Claude Code, covering its origin as an internal experiment at Anthropic, the 'do the simple thing first' product philosophy behind CLAUDE.md and context compaction, why it's built as a raw Unix-utility-style CLI rather than a polished IDE, and its per-user cost profile compared to competitors.
Key takeaways from Boris Cherny on building Claude Code
WorkOS blog recap of Boris Cherny's Acquired Unplugged interview, highlighting how Claude Code went from writing 10-20% of his code to him uninstalling his IDE entirely, Anthropic's internal engineering-productivity gains since adopting it, and Boris's claim that even durable 'human edges' like product taste may not stay durable much longer.
How I Use Claude Code
Boris Tane's practitioner blog post laying out a research-plan-annotate-implement workflow built around one core principle: never let Claude write code until you've reviewed and approved a written plan.md file, with an 'annotation cycle' where he adds inline notes directly into the plan document rather than steering through chat.
Building Claude Code with Boris Cherny
Gergely Orosz's Pragmatic Engineer interview with Boris Cherny (creator and Head of Claude Code) covering how Boris ships 20-30 PRs a day by running 5 parallel Claude instances, why plain grep/glob-driven 'agentic search' beat RAG and vector-DB approaches, how Claude Cowork was built in about 10 days, and why PRDs have effectively been replaced by working prototypes on the Claude Code team.
Using Claude Code: The Unreasonable Effectiveness of HTML
Simon Willison's link-blog post reacting to Thariq Shihipar's piece arguing for HTML over Markdown as Claude Code's output format, since HTML lets Claude drop in SVG diagrams, interactive widgets, and in-page navigation. Willison notes this changed his own default (set since the GPT-4 8k-token era) and shares his own experiment asking GPT-5.5 to generate a rich interactive HTML explainer of a real Linux security exploit.
How I Use Claude Code
Philipp Spiess's practitioner blog post sharing daily-driver patterns after subscribing to Claude Max, including calling /clear aggressively between unrelated tasks, spawning parallel sub-tasks via the Task tool to review the same code from multiple personas, using git worktrees to run Claude on two problems in the same repo at once, and being candid about where autonomous feedback loops (tests, screenshots) haven't worked well for him.
How I use Claude Code to accelerate my software engineering job and improve my life
A DEV Community practitioner post on nearly a year of daily Claude Code use, covering habits like sending agents to investigate whole services and write persistent AI_DOCS/ markdown guides, giving Claude CLI access to GitHub, AWS, Datadog, and Slack/Atlassian MCP servers, turning meeting recordings into LLM-consumable documentation via ffmpeg frame extraction, and building a personal CLAUDE.md profile to get financial and life-planning advice.
Start reading, not hoarding.
Import this vault to Burn 451 and actually read what matters.
New to Burn? See how the read-later app works →
Frequently asked questions
Who is Claude Code?
Claude Code is covered in this Burn 451 vault with a focus on agentic coding with claude. Official best practices, creator interviews, and real-world workflows for Anthropic's agentic coding CLI, Claude Code.
How was the Claude Code vault curated?
The Claude Code vault was hand-curated by the Burn 451 editorial team from publicly available essays, blog posts, podcast transcripts, and social threads. Each piece includes an AI-generated summary so readers can triage in seconds. The vault auto-syncs as new content from Claude Code is published.
How many articles are in the Claude Code vault?
The Claude Code vault currently contains 12 curated pieces organized by topic, not chronology. Each article has an AI summary and a direct link to the original source. Items are refreshed hourly through Burn 451's ISR pipeline, so new publications appear within a day.
How do I use this vault with Claude or Cursor?
Install the burn-mcp-server package from npm and connect it to Claude, Cursor, or any MCP-compatible AI tool. The vault becomes queryable as live context — your AI can search, summarize, and cite articles from Claude Code directly in conversation without manual copy-paste or re-uploading files.
What is Burn 451?
Burn 451 is a read-later app built around a 24-hour burn timer that forces daily triage. Articles you save must be read, vaulted, or released within 24 hours. The Vault layer — including this Claude Code collection — holds permanent curated reading lists for AI thought leaders, founders, and researchers.
Content attributed to original authors. Burn 451 curates publicly available writing as a reading index. For removal requests, contact @hawking520.