The Code Newsletter

Claude Code tip: put this rule in your AGENTS or CLAUDE file, and from then on Opus 5.5 𝘁𝘂𝗿𝗻𝘀 𝗯𝗮𝗿𝗲 𝗻𝘂𝗺𝗯𝗲𝗿𝘀 𝗶𝗻𝘁𝗼 𝗰𝗵𝗮𝗿𝘁𝘀 𝘆𝗼𝘂 𝗰𝗮𝗻 𝗰𝗹𝗶𝗰𝗸 𝗼𝗽𝗲𝗻 to see the trend. Stat cards, dashboards, profile pages: all of them get the upgrade. Send this to Claude Code and it adds the rule to your project's instructions 👇 "Add the rule below to this project's instructions: the CLAUDE file if there is one, otherwise the AGENTS file. Then show me what you added. ## Numbers in the UI When a screen shows stats (stat cards, profile pages, dashboards): - Many numbers of different kinds: split them into tabs - Write each one as label + value + one helper line (a rate, a comparison or a time range). Compute helpers from existing data only; if you can't, ask me - Make the key few numbers clickable tabs that open a small chart: bars or a line, plus an average line - Chart only real history. Under 4 data points: keep the number, never invent sample data - Zero or no data: show an empty state - Reuse the project's existing tab and chart components, no new dependencies"

Tell Claude Code to "make it faster" and it'll probably 𝗮𝗱𝗱 𝗮 𝗰𝗮𝗰𝗵𝗲. People who actually do perf work 𝗻𝗮𝗺𝗲 𝘁𝗵𝗲 𝗲𝘅𝗮𝗰𝘁 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲 𝘁𝗵𝗲𝘆 𝘄𝗮𝗻𝘁 instead: "Do a performance review of this project: 1. Run a profile first. Find the 5 slowest spots and how much time each one takes. 2. Check each spot against the cases below and only apply the one that matches: - It uses a cache: caching is cope for a slow design. Don't add more caches. See if it can be redesigned to be fast on its own. List the existing caches and flag which ones are hiding a slow design. - A hot loop keeps hitting hash tables: can you replace those hash tables with a flat array? Can we make the pointers an index into the array? Can we pack the most commonly used data into upper bits of the index? - Lots of objects, but each pass only reads one or two fields: should this be a structure of arrays instead of an array of structures (AoS → SoA)? - Independent awaits running one after another: run them in parallel. - Nothing matches: how do we rewrite this to better leverage data-oriented design principles? 3. For each spot, write down how much time it takes now, which case it matched, which file and lines to change, and the expected speedup. Give me the report first. Don't change anything."

try pstack! it’s my personal set of skills for rigorous and autonomous engineering with agents. it’s updated frequently as I continue to tune and simplify them 10 second tutorial: 1. run /setup-pstack and choose your budget 2. add /poteto-mode in front of any prompt you want done rigorously 3. ask /poteto-help whenever you’re unsure what pstack skills to use 4. that’s it! Grok Bot: x.ai/bot/plugin/9717366 Cursor: cursor.com/marketplace/curso… Source: github.com/cursor/plugins/tr…

Every time I have an idea, I throw it into Claude Code and let the Brainstorming skill kick in We start iterating through questions, exploring different approaches, discussing trade-offs, and refining the idea The skill helps me clarify requirements, break down the design, and validate each part before documenting the final approach By the end, I have a much clearer picture of what I actually want to build, before writing a single line of code Install: npx claude-code-templates@latest --skill development/brainstorming aitmpl.com/component/skill/d…

Like a lot of you, I use Opus 5.5 to write code and Codex Astra to review it. But Astra has a bad habit of flagging bugs that are technically possible but 𝗮𝗹𝗺𝗼𝘀𝘁 𝗻𝗲𝘃𝗲𝗿 𝗵𝗮𝗽𝗽𝗲𝗻 in practice. Opus sees the logic checks out and fixes it, then Astra flags the next one, and the loop never stops. Add this line to your review prompt, skill, or AGENTS.md: "Rate each bug from 1 to 10 on how often it would happen and how bad it would be. If the two scores multiplied are below 5, just file it as an issue and leave the code alone." Tune the threshold of 5 to your project. For Opus, the one fixing the code, add one more line to CLAUDE.md or AGENTS.md: "Do not design for rare or future cases nobody asked about."

Highly recommend adding REA to Claude Code or Codex. The name stands for 𝗥𝗲𝘃𝗲𝗿𝘀𝗲 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿 𝗔𝗻𝘆𝘁𝗵𝗶𝗻𝗴. See a feature you like in another app? Have your agent take it apart to see how it's built, then 𝗯𝘂𝗶𝗹𝗱 𝘆𝗼𝘂𝗿 𝗼𝘄𝗻 𝘃𝗲𝗿𝘀𝗶𝗼𝗻 in your project. It's been #1 on GitHub's daily trending two days running and is past 60K stars. The apps on your computer are finished products. Think of a baked cake: you can taste it, but you can't see the recipe. REA hands your agent the tools made for taking software apart. It works down layer by layer, shows you where each finding comes from, and then walks you through the recipe. You can point it at desktop apps, websites and Android apps. Ask it things like: → "How does this notes app show search results the moment I start typing?" → "Why does text copied from Notion keep its formatting when I paste it somewhere else?" → "How does this old game work out that ring of bullets?" → "Our company's 10-year-old internal tool lost its source code years ago. What is it actually doing?" Take apart your own software and open-source code as much as you like. For anyone else's, check the terms first, since many forbid reverse engineering. Don't bring it cracking or license-bypass jobs; the model will most likely refuse anyway. Run npx rea-agents setup in your terminal, pick your agent when it asks, and restart it. Then send this prompt to Opus 5.5 👇 "I want to use REA to figure out how a feature in some software works. 1. First ask me which app and which feature I want to study, then have me say whether it's my own, open source, or software whose terms allow reverse engineering. If I can't say, don't start. 2. Work out what kind of program it is first. If you need Hopper, Ghidra or IDA, tell me which one is missing and where to get it. Don't install it yourself. 3. Back every finding with evidence: which file, which function or address. Keep what you saw separate from what you're guessing. At the end, explain it all in words a 10-year-old could follow. 4. Once it's clear, tell me the minimum pieces I'd need to write to build something similar in my own project, without copying its code. Show me the plan first, and only start writing after I agree." github.com/morluto/rea

New auto-mode-setup command in Claude Code v2.1.304 This command analyzes your project and recent sessions to configure Auto Mode based on your development environment In the video, I show how simple the entire setup process is. At the end, Claude asks a few questions about your preferences to customize Auto Mode Just run /auto-mode-setup and follow the steps 👇

you can make almost any program self-improve btw step 1: prompt your agent "help me plan comprehensive logging such that we can replay what the user experienced, meaningful internal operations, and visible outcomes" step 2: setup a daily loop with this prompt "analyze my logs from this program in the past 24h thoroughly, identify broken behaviors, user friction, inefficiencies, and propose improvements sorted by estimated impact" step 3: watch your code get better every day if your program collect telemetry from users, you can do aggregated analysis which will give you even better results

Introducing Turn Inspector for Claude Code! A new Mod that lets you inspect exactly what's inside Claude's context window at every turn - System prompt - Tools - Memory - Previous turns - Model output - Tool results Watch Claude's context window fill up turn by turn and finally understand where all those tokens in your session are going Install: npx claude-code-templates@latest --mod ui/turn-inspector I love building Mods! I'm still customizing Claude Code using everything I've learned over the past 3 years building AI tools Leave a star on GitHub if you find it useful 👇 github.com/davila7/claude-co…