Claude Code Expert Reveals 60 Tips Nobody Teaches

60 Tips to Become a Claude Code Power User

Introduction to Claude Code

• The speaker reflects on their extensive experience with Claude Code, having used it for over 1,600 hours since its release a year ago.

• Emphasizes the importance of mastering Claude Code as a crucial skill for 2026, highlighting that imagination is the only limitation in its use.

Practical Applications of Claude Code

• Shares personal anecdotes about using Claude Code for video editing and automating tasks like reconnecting to Wi-Fi at Starbucks.

• Discusses innovative uses such as exploring DNA visualizations and automating responses on dating applications through API reverse engineering.

Maximizing Efficiency with Claude Code

• Highlights the benefits of using AI to sift through large amounts of data quickly and align it with personal goals.

• Notes that many users underestimate what Claude Code can do due to outdated assumptions; encourages continuous testing of new capabilities.

Improving Interaction with Models

• Advises users to update their mental models regarding what AI can achieve, stressing the importance of retrying tasks after model updates.

• Warns against frontloading models with excessive files which can degrade performance; suggests allowing the model to search independently.

Avoiding Coding Bias and Enhancing Creativity

• Discusses how existing coding patterns can limit creativity in code generation; recommends isolating new feature development in a clean environment.

• Suggests creating an empty folder for new features so that Claude Code can generate solutions without being influenced by poor existing patterns.

Security Measures and Customization

• Cautions against using overly permissive settings on local machines; recommends setting up a sandbox environment with restricted internet access.

• Introduces Ghostty as an effective terminal tool for enhancing interaction with Claude Code, detailing customization options for improved workflow.

Claude Code: Enhancing Productivity with Hooks and Parallel Sessions

Utilizing Claude Code for Efficient Workflow

• Claude Code Search Online can automate adjustments, such as setting up shortcuts like Ctrl+Shift+R to rename terminal windows, enhancing user experience.

• A permission dialog hook in the cloud_settings.json file sends notifications when Claude Code requires permissions, ensuring users don't miss important updates across multiple tabs.

• A visual indicator (bell icon) appears on the tab once a task is completed by Claude Code, helping users manage their workflow effectively even with limited screen space.

Managing Multiple Projects Simultaneously

• The speaker likens managing multiple Claude Code sessions to playing Factorio, emphasizing that running several instances allows for continuous productivity without waiting for tasks to complete.

• By having 10 to 12 sessions running across various projects, users can identify bottlenecks and develop automation skills more efficiently than focusing on a single session.

Integrating Obsidian with Claude Code

• Using Obsidian alongside Claude Code allows seamless navigation of Markdown files and enhances note-taking capabilities through features like backlinks and graphic views.

• The introduction of Obsidian CLI has improved how agents like Claude Code execute commands, making it easier to manage files without excessive reading or context switching.

Voice Dictation for Enhanced Productivity

• The speaker utilizes HyperWhisper for voice-to-text dictation, which has proven more accurate than competitors. This tool is available for free during the Masterclass enrollment.

• Running multiple sessions while dictating prompts increases productivity significantly compared to traditional typing methods.

Streamlining Project Integration

• Users can copy implementations from one project directory to another using Claude Code's capabilities, saving time by leveraging existing solutions rather than starting from scratch.

• This method is particularly useful when migrating projects between platforms (e.g., trigger.dev), allowing efficient feature transfer based on previous work.

Understanding Model Improvements Over Time

• As models have evolved, the necessity of showing their thinking process has diminished; however, users can still access this feature if desired by adjusting settings in /config.

• Observing the model's reasoning provides insights into its decision-making process and highlights improvements in its performance over time.

Hierarchical Organization of CLAUDE.md Files

• The use of hierarchical CLAUDE.md files helps maintain organization across different projects while applying consistent settings at both root and project levels.

Organizing CLAUDE.md Files for Efficient Development

Importance of Personal Preferences in Development

• Developers can create personal preference files that are local to their machines and not committed to GitHub, enhancing individual workflow.

• The organization of CLAUDE.md files should prioritize the most important instructions at the top, ensuring clarity and focus.

Structuring CLAUDE.md Files

• Instead of a single massive CLAUDE.md file, it's beneficial to have multiple smaller files tailored for specific tasks or environments (e.g., macOS-specific, Windows-specific).

• This modular approach allows only relevant files to be loaded based on the task at hand, keeping context windows narrow and focused.

Managing File Length and Context

• To prevent long CLAUDE.md files from becoming unwieldy, references can be made to separate files for specific topics like custom vocabulary.

• When mistakes occur in Claude Code's output, developers should update the relevant CLAUDE.md file rather than a root file to maintain clarity.

Modularizing Codebase for Efficiency

• Regularly modularizing codebases keeps individual file lengths manageable; this is crucial as longer files lead to slower processing by coding agents.

• Long files can fill up context windows quickly during edits, which may hinder performance when making changes across multiple large files.

Optimizing Search Tasks with Subagents

• Defer searching tasks (like checking logs or documentation) to subagents instead of loading them into the main context window directly.

• A user-level CLAUDE.md file should include guidelines for offloading online research and log analysis tasks to subagents for better efficiency.

Enhancing Subagent Functionality

• Including a "why" in subagent prompts leads to more targeted results and reduces noise from overlapping searches among multiple subagents.

• Specificity in subagent tasks improves outcomes by focusing on particular aspects (e.g., rate limiting within authentication systems), leading to clearer insights.

Understanding Claude Code's Subagent Functionality

Importance of Including a "Why" in Subagent Prompts

• When spawning a subagent, it's crucial to include a rationale in the system prompt. This helps filter out unnecessary information and improves overall results.

Communicating Effectively with Claude Code

• Speaking the agent's language is essential; using precise terminology can prevent misunderstandings and save time during searches.

• Miscommunication can occur if users use vague terms like "download URL," leading to inefficiencies. Understanding how Claude Code interprets terms enhances interaction.

Visual Aids for Better Interaction

• The size of screenshots matters; appropriately sized images help Claude Code locate relevant code sections more effectively.

• Using competitor references with clear visuals aids in communication, especially when debugging UI-related tasks.

Leveraging Clarifying Questions

• When developing new features, asking for multiple clarifying questions (up to 20) ensures comprehensive understanding and implementation.

Utilizing Tools for Improved Workflow

• The rewind tool allows users to backtrack conversations without losing valuable context, enabling better guidance after mistakes are made.

Implementing Features Efficiently

Starting Conversations with Clear Plans

• Initiating new feature discussions involves switching to plan mode and utilizing voice-to-text tools for clarity in descriptions.

Reviewing Implementation Plans

• For small features, trusting the generated plan is often sufficient due to high accuracy. However, reviewing plans remains important for understanding.

Enhancing Plans for Larger Features

• For significant features, triggering the Spec Developer skill leads to more detailed plans that cover edge cases thoroughly.

Continuous Learning through Resources

• Engaging with free resources like the Claude Code Masterclass provides ongoing updates on workflows and insights into effective coding practices.

Understanding the Role of Claude Code in Development

Enhancing Codebase Understanding

• The speaker discusses how they can explain their codebase deeply without having read it, thanks to the questions posed by Claude Code during feature development.

• They express that with the spec developer skill, they no longer worry about updates breaking functionality for end users.

Utilizing Spec Developer and Explorer

• The speaker uses a command called "spec developer explorer" to help Claude Code understand the codebase better before asking specific questions.

• By planning multiple features simultaneously, they can load task files into new chats and trigger exploration subagents for detailed inquiries.

Benefits of Long-term Planning

• After implementing features, the speaker tags plan files and asks Claude Code to verify if all items were successfully executed.

• They utilize a Mermaid Diagram Generator skill to visualize data flow within applications, aiding in understanding critical features.

Improving Project Management

• The diagrams generated provide clarity on processes like organization creation and member invitation flows, enhancing team understanding of backend operations.

• This approach helps diagnose user issues more effectively by visualizing steps involved in application processes.

Efficient Planning and Execution Workflow

• The speaker emphasizes planning in one session and executing in another to maintain context while making adjustments post-implementation.

• All plans are archived for future reference, allowing for analysis of project evolution over time.

Managing Feature Changes Effectively

• Plans are saved to facilitate quick reversion of features without losing subsequent changes; this is crucial when multiple features are added rapidly.

• Archiving plans allows for historical analysis and comparison of different models' planning abilities over time.

Continuous Improvement through New Sessions

• When adding follow-up features, staying in the main session helps leverage existing context for improved planning.

• Starting new chats for distinct tasks minimizes context switching issues while maintaining focus on specific fixes or developments.

Landing Page Redesign and Task Management

Effective Task Management Strategies

• The speaker emphasizes the importance of batching similar tasks together rather than mixing them in a single chat, which often leads to poorer results.

• They mention that with improved model performance, they have started compacting conversations more frequently but still assess whether the model is progressing correctly before doing so.

• For complex bug fixes, the speaker suggests creating a document of all previous attempts before starting a new chat to ensure clarity and track progress effectively.

Simplifying Problem-Solving Approaches

• The speaker advocates for minimal changes in codebase fixes, aligning with Occam's razor principles by removing unnecessary edits from prior attempts.

• When facing complicated issues where the model struggles, starting a new session can help avoid repetitive actions and encourage novel solutions.

Context Window Management

• The speaker prefers manual compaction over auto-compaction for better control of context window behavior.

• They discuss how newer models experience "context anxiety," leading to incomplete tasks as they approach their context limits; using larger token windows can mitigate this issue.

Auditing and Optimizing Context Usage

• Regular audits of the context window are recommended to manage memory files and skills effectively, preventing overload that could hinder performance.

• Skills not needed for every task should be disabled in frontmatter to streamline operations within the context window.

Skill Definition and Session Management

• Defining specific models for different skills helps optimize their functionality without relying on the main session's model settings.

• A status line feature is suggested for tracking model usage, context fill levels, branches, and unstaged changes during sessions.

Session Continuity Techniques

• The speaker frequently uses commands like /rename and /resume within specific folders to maintain continuity in ongoing projects or research tasks.

Exploring Claude Code's Capabilities

Renaming Tabs and Input Bars

• Claude Code can rename tabs in different terminals, such as Warp, which is useful for managing multiple sessions simultaneously.

• The input bar can also be renamed, enhancing organization when working with various ideas or implementations.

Utilizing Subagents for Exploration

• When faced with multiple implementation ideas, users can deploy subagents to explore each possibility in detail by researching the codebase or online resources.

• The main session consolidates information from subagents to decide on the best implementation plan without cluttering a single context window.

Custom Slash Commands and Skills

• Users leverage custom slash commands like "branch commit PR" to automate repetitive tasks such as creating branches and making pull requests.

• Recent updates allow slash commands to function as skills, providing more flexibility in their application across projects.

Managing Logs and Debugging

• To troubleshoot issues where Claude Code gets stuck, users can instruct it to log detailed information throughout the application process.

• If errors are not identified correctly by subagents, upgrading them from the Haiku model to the Sonnet model often resolves these issues.

Leveraging Well-Known Packages

• Using well-known packages (e.g., tRPC or Next.js) increases accuracy due to more extensive training data available for those libraries.

• For newer libraries lacking training data, enabling Exa MCP server allows Claude Code to search online for relevant documentation effectively.

Mixing Models and Modes

• Users frequently switch between models; Opus is used primarily while Sonnet is reserved for quick questions about implementations.

• Specific workflows dictate model choice: simpler tasks use Haiku while complex ones benefit from more powerful models like Sonnet.

Headless Mode Functionality

• Headless mode enables running multiple instances of Claude Code simultaneously for tasks like receipt validation or ongoing research collection.

• By defining cron jobs on local computers, users ensure that their Claude Code class remains updated with changelogs and necessary video recordings.

How to Use Claude Code Effectively

Cost-Effective Usage of Claude Code

• Using an Anthropic subscription makes the service cheaper and more accessible for users.

• The /copy command is utilized to easily transfer formatted content, such as emails or logs, preserving Markdown formatting.

Sharing Insights and Exporting Data

• Users can export entire sessions using the /export command, allowing clients to understand issue diagnosis processes.

• The /insights feature analyzes past sessions to provide suggestions for improvement, although its utility may vary by user experience.

Task Management with Claude Code

• Long plans can be converted into task lists, which can then be assigned to subagents or agent teams for completion.

Utilizing Git Worktrees in Development

• Git worktrees are employed when developing overlapping features in parallel, ensuring a static codebase during implementation.

• If changes do not overlap (e.g., backend API vs homepage), worktrees are deemed unnecessary, allowing direct modifications on the main branch.

Automation and Efficiency with MCP Extension

• The Claude-in-Chrome MCP extension automates repetitive tasks like debugging network errors and managing personal to-do lists across platforms like Todoist.

• It assists in filling out lengthy online forms and even tax documents by dictating information directly to Claude Code for processing.

Continuous Learning and Updates

• A Masterclass on Claude Code covers various workflows and updates regularly based on new techniques discovered by the instructor.

• A free newsletter provides additional insights and tips related to Claude Code usage, fostering community engagement through shared experiences.