Claude Code First Prompt
Your first Claude Code prompt is a high-fit primary-source video because it shows the agent transition at the exact moment of delegation. The tutorial tells a new Claude Code user to write a descriptive request, decide whether file changes should be accepted automatically or require explicit approval, use plan mode for read-only codebase research, review the plan, and inspect the resulting change. The example is deliberately ordinary: adding a dark-mode toggle with contrast choices that fit an existing light theme.
The strongest Spiralist relevance is the first boundary around agency. A beginner prompt is not just a request for code; it sets scope, permissions, review expectations, and the user's relationship to the agent's work. Plan mode matters because it separates investigation from mutation, while explicit approval modes keep the user aware that an assistant is about to alter a real system. That belongs beside AI Coding Agents, AI Agents, Agent Tool Permission Protocol, Agent Audit and Incident Review, Context Windows and Context Engineering, and Humane Friction Standard.
External sources support the product pattern while limiting the claim. Anthropic's common workflows documentation describes plan mode as read-only codebase analysis for exploration, complex-change planning, and safe review before edits. Anthropic's Claude Code best-practices guide recommends asking Claude to make a plan before coding, keeping project context close to the repository, and managing permissions deliberately. NIST's AI Agent Standards Initiative supplies independent policy context for why agent identity, authorization, secure operation, interoperability, and evaluation matter as software agents begin acting for users.
The limits should stay clear. This is an official Claude product tutorial from May 15, 2026, not an independent evaluation of accuracy, defect rates, security, accessibility, or real-world developer productivity. It is strong evidence for Anthropic's preferred novice workflow and vocabulary around prompts, file-change approvals, command permissions, and plan mode. It does not prove that first-time users will notice bad plans, that auto-accepted edits are harmless, or that a good-looking UI change has satisfied maintainability, contrast, testing, or product requirements.