ZKidz Dev LLC

ZKidz Dev LLC No fluff, no ghost devs — just a team that shows up and ships on time.

We build apps people actually love. 8+ years shipping e-commerce, marketplace & custom software for founders worldwide. 300+ storefronts, 15+ apps, 99% success rate.

Quick Claude Code tip: match the model to the task with `/model`, and your pick sticks until you change it.Rough guide: ...
08/02/2026

Quick Claude Code tip: match the model to the task with `/model`, and your pick sticks until you change it.

Rough guide: Opus for hard reasoning, Sonnet for balanced everyday work, Haiku for fast and cheap. You can switch mid-session as the work changes.

Worth trying: the `opusplan` alias uses Opus while you plan, then switches to Sonnet to execute, so you get strong plans without paying Opus rates for the whole build.

The model matters, but how you write the prompt matters just as much. These habits work on Claude and carry over to othe...
08/01/2026

The model matters, but how you write the prompt matters just as much. These habits work on Claude and carry over to other AI tools.

Be clear and direct: say the goal, the audience, and the format. Show two or three examples of the output you want, since examples steer tone better than a long description. Use tags like `` and `` to keep your instructions separate from your data. And ask the model to think before it answers.

Shortcut: the Claude Console has a prompt generator and a prompt improver to draft and refine prompts for you.

Small Claude Code habit that pays off: use `@` to hand it exactly what to look at.`/auth.ts` drops that file's contents ...
07/31/2026

Small Claude Code habit that pays off: use `@` to hand it exactly what to look at.

`/auth.ts` drops that file's contents into the prompt, so Claude reads the real code instead of searching for it. If you have an MCP server connected, `:resource` pulls a doc or ticket from that tool, and the server's prompts show up in the `/` menu to run like commands.

Quick win: mentioning a file is faster than asking Claude to find it. It skips the search and reads the file on the first try.

Claude Code can run without you. A routine is a saved prompt plus its repos and connectors, running in the cloud with yo...
07/30/2026

Claude Code can run without you. A routine is a saved prompt plus its repos and connectors, running in the cloud with your laptop closed.

It can trigger on a schedule, on an API call from a tool like your monitor, or on a GitHub event like a new pull request, and you can combine those. Create one with `/schedule daily PR review at 9am` or the web form, then review each run as a full session.

Heads up: routines are in research preview, so limits can change and runs count against your daily allowance.

Tired of re-prompting Claude Code for the same tone every time? Set an output style once.It changes how Claude responds ...
07/29/2026

Tired of re-prompting Claude Code for the same tone every time? Set an output style once.

It changes how Claude responds by editing the system prompt. Run `/config`, open Output style, and pick a built-in: Explanatory adds teaching notes, Learning has you fill in `TODO(human)` code, Proactive just acts. You can also write your own as a markdown file in `.claude/output-styles/`.

Good to know: a custom style drops the built-in coding instructions unless you set `keep-coding-instructions: true`. Keep it on when you still want Claude coding, just in a different voice.

Handy Claude Code safety net: it saves a checkpoint before every edit, so you can undo fast.Open the menu with `/rewind`...
07/28/2026

Handy Claude Code safety net: it saves a checkpoint before every edit, so you can undo fast.

Open the menu with `/rewind` or a double Esc on an empty prompt. It lists your prompts from this session, you pick a point, and you choose what to bring back: the code, the conversation, or both. The same menu can also summarize part of the chat to free up context.

Worth remembering: checkpoints cover Claude's edits, not files changed by bash commands, and they do not replace Git. Treat them as local undo.

Quick Claude Code tip: if it starts forgetting earlier detail, the context window is full. Three commands fix that.`/con...
07/27/2026

Quick Claude Code tip: if it starts forgetting earlier detail, the context window is full. Three commands fix that.

`/context` shows what is filling the window and what to trim. `/compact` summarizes the chat so far, and you can add a focus like `/compact focus on the API work` to keep what matters. `/clear` wipes the conversation when you switch tasks, and the old session stays saved on disk. Claude also compacts on its own near the limit.

One worth knowing: a line in CLAUDE.md can tell Claude what to preserve when it compacts, like the modified files and your test commands.

Claude Code is good on its own, but it doesn't know your issue tracker or your database until you connect them. That is ...
07/17/2026

Claude Code is good on its own, but it doesn't know your issue tracker or your database until you connect them. That is what MCP servers are for.

Adding one is a single command. For a hosted server, run `claude mcp add --transport http ` and check it with `claude mcp list`. If it needs a login, finish the sign-in from the `/mcp` panel. Local tools work the same way, with everything after `--` being the command Claude runs, so `claude mcp add playwright -- npx -y /mcp@latest` hands Claude a browser it can drive.

Want your teammates to have it too? Add it with `--scope project`, commit the `.mcp.json` it writes, and everyone who clones the repo gets it after one approval.

Quick Claude Code tip: stop reminding it to run the formatter or leave your secrets alone. Wire it into a hook.A hook is...
07/16/2026

Quick Claude Code tip: stop reminding it to run the formatter or leave your secrets alone. Wire it into a hook.

A hook is a shell command that runs at a set point in Claude Code's lifecycle. The name says when it fires: a `PreToolUse` hook runs before Claude uses a tool, a `PostToolUse` hook right after. So a `PostToolUse` hook can run Prettier after every edit, and a `PreToolUse` hook can block an edit to `.env` and hand Claude the reason. Add them to a settings file and run `/hooks` to see what you have.

One worth stealing: set a hook's `type` to `prompt` and a Claude model makes the call. A `Stop` hook can check whether the work is really finished and push Claude to keep going if not.

Quick Claude Code tip: make Claude plan before it edits.Press Shift+Tab or start a prompt with /plan and Claude switches...
07/15/2026

Quick Claude Code tip: make Claude plan before it edits.

Press Shift+Tab or start a prompt with /plan and Claude switches to plan mode. It reads your code and writes out a plan without changing a thing. When you like the plan, you approve it and the edits start. When you don't, you keep talking until it's right.

Worth stealing: press Ctrl+G to open the plan in your own editor and rewrite the steps before Claude runs them.

Address

30 North Gould Street
Sheridan, WY
82801

Alerts

Be the first to know and let us send you an email when ZKidz Dev LLC posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Shortcuts

Share