, I’ll cover my experience on how you may get essentially the most out of Claude Code. Claude Code is a strong coding command-line interface (CLI) tool. You possibly can open it directly in your terminal and begin coding.
This makes it an efficient tool each for proficient coders acquainted with a terminal. Nonetheless, it also works well for non-technical people, since they’ll code just by typing natural language, and don’t need to interact with an advanced IDE.
In this text, I’ll cover why it is best to think about using Claude Code for coding, highlighting its effectiveness at implementing code with little to no manual review. Moreover, I’ll cover some specific techniques I utilize to get essentially the most out of Claude Code, before highlighting some limitations of Claude Code.
I’m not affiliated with Claude Code, and I’m simply writing this text from my experience using the tool.
Why use Claude Code
If you need to practice coding nowadays, it is best to embrace agentic coding. I feel this is applicable whether you only began programming or when you’ve been programming for 10+ years. All of us must be leveraging coding agents to turn into more efficient engineers.
Now you’ve gotten a big number of such tools to pick from. Claude Code is one in all several CLI tools, with another options being OpenAI Codex or Gemini CLI. These CLI tools are good when you don’t need to take a look at the code you’re developing, typically for less complicated implementations or bug fixes.
Nonetheless, I’m noticing increasingly more that my coding agents are in a position to implement larger scale implementation with fewer and fewer errors, which is noteworthy. I consider we’ll soon be in an era where coding agents are doing all coding (you would also argue we’re already there), and in such a scenario, CLI tools can be powerful, because you never have the necessity to take a look at any code.
In case you do need to look more on the code, nonetheless, you may consider more integrated agentic coding experiences, in tools comparable to Cursor or Antigravity, that are forks of VS Code, but with agentic functionality deeply integrated into the appliance.
I like using each CLI’s and IDEs for agentic coding. If I would like to code without performing manual reviews of the code, I are inclined to use Claude Code increasingly more. I feel scenarios where it is best to strongly consider utilizing Claude Code are:
- You’re fixing a bug
- You’re investigating production logs
- You’re implementing a simple or medium difficulty feature
In case you’re in any of those situations, or similar, I strongly urge you to check out Claude Code, or some other CLI coding tool, and see how well it performs.
Techniques I take advantage of
On this section, I’ll cover a number of the techniques I utilize to get essentially the most out of Claude Code. Each subsection will cover a selected technique, including why I feel it’s vital.
Slash commands
Slash commands are prompts that you just store for quick access. You usually do that on a prompt you run regularly, comparable to:
- Investigating production logs
- Making a PR
- Checking if code is production-ready
The moment I realize I’m running a prompt for the second time, I create a slash command with the prompt. This makes it way more efficient to run the prompt in the long run, since I can simply call the command starting with a slash (/), as a substitute of typing out the prompt each time.
The primary reason for using slash commands is time savings and consistency. Writing out prompts takes time, after all, and moreover, you’ll often be inconsistent when writing out the identical prompt multiple times.
For instance, if you’ve gotten a prompt for checking if code is production-ready, you would possibly have a checklist the agent should undergo, comparable to:
- Searching for missing types
- Running pre-commit checks
- Spinning up the Docker image
Nonetheless, when you’re typing out the prompt each time, you risk forgetting any of those points, which then makes the agent less effective.
User memory and project memory
I’m also a heavy user of each the user memory and project memory in Claude Code. User memory is the memory that applies to all your Claude code sessions, while project memory is the memory that only applies to the folder and subfolders you’re currently working in.
I utilize the user memory each time I would like Claude to recollect specific preferences I even have that apply across repositories. That is for instance:
- Learn how to create good PR’s
- Learn how to check production logs
- Details about my IaC stack
If I even have preferences that only apply inside a selected project (or a minimum of that don’t apply to all my coding projects), I store them within the project memory, in a file called AGENTS.MD. I are inclined to use AGENTS.md as a substitute of CLAUDE.md, because then all my coding agents will find the file, for instance, if I determine to try Cursor or other tools in the long run.
Utilizing memory is essential, because it prevents Claude Code from making the identical mistake twice. If Claude used the flawed table name or checked the flawed log group, you may store in its memory which mistake it made, and the proper table name and log group name. While you interact with Claude Code in the long run, it is going to then remember this information and be more efficient next time.
In case you proceed doing this for all of the mistakes Claude Code makes, it is going to be personalized to your codebase and preferences and can be far simpler for you as a coding agent.
Plan mode
Plan mode can be incredibly vital when using Claude Code. Plan mode is actually the paradox killer. While you prompt Claude Code in plan mode, it is going to analyze your prompt and codebase and ask questions on anything that’s unclear or ambiguous.
Plan mode is the paradox killer
That is super helpful, because it’s near unimaginable to jot down a very unambiguous prompt, and having Claude Code discover unclear parts of your prompt and ask you questions on it’s a superb method to align on how a coding implementation must be done.
I probably start 90% of my Claude Code sessions with plan mode; the one exceptions are after I’m performing some super easy fixes or just writing experimentation code. I like to recommend trying plan mode as much as possible, and only not using it while you truly feel it’s not obligatory.
Do every thing with Claude
My last tip is that only your imagination limits what you may do with Claude Code. For instance:
- Do you need to make a presentation? Ask Claude to make a presentation in Latex Code
- Do you need to analyze CloudWatch logs? Give Claude AWS permissions and have it analyze for you
- Do you need to construct a portfolio website? Have Claude vibe it up and deploy it to Vercel. You don’t even have to go away the terminal
The probabilities are infinite, and I strongly urge you to make use of the next mindset:
Every time you encounter an issue, it is best to think:
This mindset will prevent quite a lot of time, because you’ll realize that quite a lot of tasks may be solved by simply providing Claude with the required permissions, and let it try to solve the issue for you.
Challenges with Claude Code
I’ve mentioned quite a lot of positive sides of Claude Code in this text, since it’s a extremely good tool that makes me far simpler as an engineer. Nonetheless, there are also some downsides to Claude Code.
The primary issue I even have with Claude Code is that it’s a bit slow sometimes, especially after I compare it to using Cursor. This happens although I’m using Claude Sonnet 4.5 or Claude Opus 4.5 in Cursor.
I’m not entirely sure why that is the case, but I often find that when analyzing log groups or performing easy fixes, Claude in Cursor is just faster than using Claude Code.
One other downside to Claude Code is that it’s purely terminal-based. If you need to understand your code higher and review it, I’d encourage you to make use of an IDE as a substitute. It also needs to be noted that there exists a Claude Code extension for VS Code, which may be installed in any VS Code fork as well, comparable to Cursor.
Conclusion
In this text, I’ve talked about Claude Code, covering why and when it is best to utilize the tool. I highlighted that Claude Code excels at coding while you don’t must perform manual reviews of the code, which is increasingly more often the case with the rapid improvements in LLMs. I then covered specific techniques I take advantage of to get essentially the most out of Claude, comparable to heavy usage of memory and plan mode, in addition to attempting to apply Claude Code to just about every problem I encounter. I consider coding agents are here to remain, and that it is best to embrace the technology as much as you may. It is best to always be testing the bounds of what these coding agent tools can do, with a view to turn into as effective an engineer as possible.
👉 My Free Resources
🚀 10x Your Engineering with LLMs (Free 3-Day Email Course)
📚 Get my free Vision Language Models ebook
💻 My webinar on Vision Language Models
👉 Find me on socials:
🧑💻 Get in contact
✍️ Medium
