Install with an AI agent

Open Core HR is built to be installed by AI coding agents. Every guide in this manual is published as plain Markdown, the whole manual is available as llms.txt and llms-full.txt, and the install steps are deterministic shell and Artisan commands that agents run well. There is no charge for this: it is part of your licence. Choose an installation plan only when you want our engineers to do the work, build your mobile apps or train your team.

Two ways to use this guide

• Locally, to try it out. Run Open Core HR on your own Mac, Windows or Linux machine with Laravel Herd, Docker or any local PHP stack. Nothing is exposed to the internet, you can reset it at any time, and the same agent can later move you to a server. The local development guide has the stack details.
• On a server, to go live. Install on a cPanel account, a Linux VPS or another platform for your company to use.

The prompt below works for both. Tell the agent which one you want.

What you need

• Locally: a computer that can run PHP 8.5, Composer, Node.js and a database (Laravel Herd on Mac and Windows gives you all of this), plus the release package extracted in a folder.
• On a server: a server or hosting account that meets the requirements, with SSH access (or cPanel Terminal) for the agent to run commands, and the release package extracted there.
• The release package itself, from your CZ App Studio account.
• An AI coding agent that can run shell commands where the files are: Claude Code, Cursor, Codex CLI, GitHub Copilot in VS Code, Gemini CLI or similar.
• For a server: your domain, database credentials and SMTP details. Give them to the agent through environment variables or files on the server, never by pasting them into the chat.

Step 1: Open the agent where the files are

Start the agent in the folder where you extracted the Core download (the guides call it backend/). It contains AGENTS.md and INSTALL-WITH-AI.md, which your agent reads first.

Locally, open that folder in your editor or terminal:

cd ~/Projects/open-core-hr
claude      # or: cursor . / codex / gemini

On a server, connect first and then start the agent:

ssh deploy@your-server
cd /var/www/open-core-hr
claude      # or: cursor . / codex / gemini

Step 2: Paste the install prompt

Copy this prompt into the agent. It tells the agent where the documentation is and sets the ground rules.

You are setting up Open Core HR, a self-hosted Laravel application, in this directory.
Read AGENTS.md and INSTALL-WITH-AI.md in this folder first; they describe this release.
For the full manual, use the open-core-hr MCP server if it is connected, otherwise read
https://czappstudio.com/llms-full.txt.
Follow these guides in order: Requirements, Prepare, the recipe for my target
(Local development, cPanel, Linux VPS or other), Configuration, Services, then the
Installation checklist. For a local setup, skip HTTPS, domain and production hardening
steps and say which ones you skipped.

Rules:
- Explain what each step does before you run it, and run one step at a time.
- Ask me before anything destructive (dropping databases, overwriting files, restarting services).
- Run `php artisan doctor` after each stage and fix what it reports.
- Never ask me to paste passwords, API keys or licence keys into this chat. Tell me which
  environment variable or file to fill in, then continue once I confirm.
- Only backend/public may be served over HTTPS. Verify this.
- When finished, write a deployment report in DEPLOYMENT-REPORT.md listing versions,
  the checks you ran from the Installation checklist, and anything you could not verify.

My target is: <my own computer for a trial | cPanel | Linux VPS | other>.
My domain (server only) is: <hr.example.com>.

Step 3: Let the agent work, and answer its questions

The download includes AGENTS.md, INSTALL-WITH-AI.md, ready-made skills and an MCP connection to this manual, so your agent starts with the architecture and the rules already loaded (see Using AI coding agents). The agent reads the manual, checks PHP, extensions and the database, writes .env, runs migrations, builds assets, configures the queue worker and scheduler and, on a server, sets up HTTPS and email. It will stop to ask you for credentials or decisions. Keep these rules in mind:

• Fill secrets into .env yourself when asked, then tell the agent to continue.
• If the agent proposes a command you do not understand, ask it to explain. A good agent will.
• Let it finish the installation checklist; the report it writes is your record of what was verified.
• Started locally and ready to go live? Open the agent on the server and paste the same prompt; it can also read your local DEPLOYMENT-REPORT.md and .env layout to reproduce your choices.

Step 4: Sign in and activate

Open your local address or your domain, sign in as the administrator the agent created and remove the bootstrap password from configuration as described in Configuration. Then open Settings → Modules, choose Activate product and approve on the CZ App Studio website; you return to the product activated. Activation is what lets you install the add-ons you own and receive updates. A local trial uses your licence's separate local slot. See Licences and activation.

Using AI on any page of this manual

Every page has an Use this page with AI toolbar:

• Copy as Markdown copies the page in plain Markdown to paste into any chat.
• Open in Claude and Open in ChatGPT start a conversation with the page already loaded.
• Copy prompt for your coding agent copies a short prompt that points your agent at this page.
• View as Markdown opens the raw .md file at the same address with .md appended, for example /docs/installation/requirements.md.

Agents that look for llms.txt find it at the site root.

When to ask our team instead

AI handles a standard installation well. Choose a paid installation plan when you want our engineers to install on your infrastructure, build and submit branded Android and iOS apps, hand the system over to your administrators in person or run live training. Installation plans are for people's time; the AI route stays free.

If something goes wrong

Work through Troubleshooting with your agent. If you need us, open a ticket from your account and attach the agent's DEPLOYMENT-REPORT.md with secrets removed. It tells us exactly what was done.