I have been using claude for a while, it works quite great for writing code. But sometimes I need some high level insights but it straight away starts implementing things.<p>I add some smippets in the prompt itself like "just discuss and do not write the code".<p>Is there a better way to tackle this?<p>Model - Claude 3.5 Sonnet
Whenever I start a new chat with claude for code, I give it instructions for how I want it to work with me.<p>Here's a basic example:<p>----------------------<p>Hi Claude! I'm working on[coding project]. We've made great progress, and I want to continue working with you.<p>PROJECT CONTEXT:<p>- [technologies]<p>- [more about the technologies]<p>- [what I am trying to do]<p>WORKING PREFERENCES (for me and you to work together):<p>1. DEVELOPMENT PREFERENCES<p>1.A. Strongly prefer step-by-step approaches, breaking down complex problems into smaller, manageable pieces<p>1.B. Like to verify each step before moving to the next one<p>1.C. Appreciate clear explanations of "why" we're doing something, not just "what" to do<p>1.D. Learn best through an interactive process where I can ask questions and see results<p>1.E. Value confirmation of current state before making changes (like checking [tool specific info])<p>1.F. Prefer to understand potential risks or implications before executing commands<p>1.G. Like to see the "before and after" state of changes (like checking database structure)<p>1.H. Appreciate analogies and explanations that connect to broader concepts<p>1.I. Value clear, direct communication about errors or issues<p>1.J. Learn well through troubleshooting - using errors as learning opportunities<p>1.K. Like to work on one feature at a time<p>1.L. Bias towards action and reversible decisions<p>1.M. Like clean, minimal code first<p>1.N. Try to lead with "test driven development", so write tests first<p>1.O. Enjoy testing things in locally<p>1.P. Start with a clear, single action<p>1.Q. Please include verification steps<p>1.R. Please explain what we expect to see<p>1.S. Please provide context for why something might have failed<p>1.T. Please give me space to ask questions before proceeding<p>1.U. Need to understand database state and impacts across all environments (local, test, production)<p>1.V. Prefer to see concrete examples of what success looks like before starting<p>1.W. Like to commit and verify small changes before moving to next steps<p>1.X. Value understanding both the immediate fix and the underlying concepts<p>2. TROUBLESHOOTING PREFERENCE<p>2.A. Prefer to tackle one error at a time<p>2.B. Like to understand the full error message before proceeding with fixes<p>2.C. Value seeing the database/application state before and after troubleshooting steps<p>2.D. Appreciate understanding similar issues that might arise in the future<p>3. CODE STYLE:<p>3. A. [programming language specific likes/dislikes]<p>3. B. [what to include like comments etc]<p>3. C. [what to not include]<p>3. D. * Check before generating code *<p>3. E. Tests should be as simple as possible and only test one thing<p>[4 - redacted, very specific to the project I'm working on]<p>5. COMMUNICATION STYLE<p>5.A Prefer technical explanations broken down into digestible pieces"<p>5.B Value confirmation of understanding before proceeding"<p>5.C Appreciate acknowledgment when something is complex or potentially risky"<p>5.D Like to see the reasoning behind suggested approaches"<p>6. MY LEARNING STYLE<p>6.A Learn best through iterative problem-solving"<p>6.B Value seeing the connection between different parts of the system"<p>6.C Prefer to understand both the 'happy path' and potential failure modes"<p>6.D Appreciate real-world analogies and examples"<p>What do you need from me to get started?<p>----------------------<p>Because Claude re-reads the whole chat every time it responds to a chat, once it's in the chat, it keeps all of my preferences in mind every time it responds.<p>Then when I finish with the task/chat, I ask it what it found helped work with me and what my working style was like. Then, I update the above prompt and ask if anything is missing or if anything would be helpful for the next time we start a chat.<p>You can also setup a "Project" if you are doing a specific project to serve as a "coding" umbrella if you will. Within the project definition, you can "__Set project instructions__" - which is basically a test box in which you can "Provide Claude with relevant instructions and information for chats within Learnagon. This will work alongside user preferences and selected style in a chat." This is then present within all of the chats with the project.<p>Since I update my instructions for the coding every single chat based on preferences and what I learned and Claude learned about working with me the previous go around, I find it easier to have a text file that I update and keep track of my prompt so that each new chat gets the updated on.<p>I saw a new prompting strategy this evening on Twitter/X that I haven't tried, but will try the next time I'm using Claude => <a href="https://twitter.com/DaveShapi/status/1861924788761169926" rel="nofollow">https://twitter.com/DaveShapi/status/1861924788761169926</a><p>If you try the X one, I'd love to hear your experience :)