Get started
Build with AI
Connect an AI coding agent such as Claude Code, Codex or Cursor to your project with one prompt.
How it works
CloudHike.dev is built so an AI coding agent can wire it into your app without you copying values by hand. The pieces are:
- Quick Start with AI on the project's Overview tab gives you one prompt to copy.
https://cloudhike.dev/prompt.mdis a plain Markdown instruction file written for agents. It tells them how to download your .env and use every service.- Agent token: a secret that lets the agent download only your project's .env. It is created when you first copy the prompt.
- Docs as Markdown: every docs page is available as raw Markdown at
https://cloudhike.dev/docs/md/<page>, andhttps://cloudhike.dev/llms.txtlists them all.
Step by step
- Deploy the services your app needs on the project's Services tab (see Quick start).
- Open the Overview tab and click Copy prompt. The first time, this creates an agent token.
- Open your app's folder in your AI tool and paste the prompt.
- The agent downloads the .env, adds it to your env file, checks it is git-ignored, and writes the code that connects each service.
- Review the changes it made, then run your app.
The prompt you copy looks like this (the token is filled in when you copy it):
Fetch and follow the instructions at https://cloudhike.dev/prompt.md to set up the CloudHike.dev backend in this app.
Project id: <your project id>
Access token: ic_...One prompt for your whole workspace
The Connect your coding agent card on the Coding agent page in the sidebar gives an agent your whole workspace instead of a single project. The agent can list your projects, create one, deploy services and download the .env, then connect the services to your app.
Fetch and follow the instructions at https://cloudhike.dev/workspace-prompt.md to set up CloudHike.dev as the backend for this project.
API key: icu_...- Workspace API key: created when you first copy the prompt. Use New key to replace it, or Revoke to turn it off.
- The key can create projects, deploy services and read a project's keys through
https://cloudhike.dev/api/v1. It cannot delete projects or services: that stays in the dashboard. - Everything the agent does follows your plan limits and shows up in the audit log.
Treat the key like a password. Anyone who has it can use your workspace until you revoke it.
Set up your AI tool
Any agent that can fetch a URL and run shell commands works. These are the common ones.
Claude Code
cd my-app
claude
# then paste the prompt. Or start with it directly:
claude "Fetch and follow the instructions at https://cloudhike.dev/prompt.md ..."To make it permanent, add a short note to CLAUDE.md in your repo, for example: "Backend is hosted on CloudHike.dev. Read https://cloudhike.dev/docs/md/quick-start before changing database, auth, storage or Redis code. Secrets live in .env.local and must never be committed."
Codex
cd my-app
codex
# paste the prompt, or:
codex "Fetch and follow the instructions at https://cloudhike.dev/prompt.md ..."Codex reads AGENTS.md from your repo. Put the same note there so every session starts with it.
Cursor
Open your project, start a new Agent chat and paste the prompt. For a lasting rule, add a file such as .cursor/rules/instacloud.mdc with the same note as above.
Antigravity, Windsurf, Copilot agent and others
Open the project and paste the prompt into the agent chat. If the tool cannot browse URLs, open https://cloudhike.dev/prompt.md yourself, copy its text and paste it after the prompt.
ChatGPT, Claude.ai and other chat apps
Chat apps cannot edit your files, but they can write code for you. Paste the Markdown of any docs page (use Copy page as Markdown) and say what you want to build. Add your values to the env file yourself.
The agent token
- It only downloads this project's .env. It cannot deploy, delete or change anything.
- Reset creates a new token and stops every prompt copied before.
- Revoke removes the token. Agents get a 401 until you copy the prompt again.
- Every download is written to the audit log on the Usage page.
- The prompt tells agents not to store the token in your app. Still, treat a copied prompt like a password and do not paste it into shared chats.
Tips for good results
- Deploy services first. The agent only wires up services that are ready, and tells you when one is missing.
- Name the feature. "Add sign-up and a profile page with Auth" works better than "use CloudHike.dev".
- Point it at the matching docs page, such as
https://cloudhike.dev/docs/md/storage. Every page ends with a ready-made prompt. - Ask it to use your existing ORM, UI kit and folder layout.
- Review every diff. Agents make mistakes, and you own the result.
- Check that secrets are not printed, committed or used in browser code.
Use with AI
This is the page to give your agent first. It explains the whole flow in one place.
- 1
Open a terminal in your app's folder and start Claude Code.
claude
- 2
Paste the prompt below and press Enter. Approve the file edits and commands it asks about.
- 3
Optional: save a short note about CloudHike.dev in CLAUDE.md so every session knows about it.
Read https://cloudhike.dev/docs/md/build-with-ai and https://cloudhike.dev/prompt.md, then set up CloudHike.dev in this app. Project id: <paste from the Overview tab> Access token: <paste from the Overview tab> Also add a short note to this repo's agent instructions file (CLAUDE.md, AGENTS.md or Cursor rules, whichever exists) saying the backend is hosted on CloudHike.dev, secrets live in the env file and must never be committed, and that docs are at https://cloudhike.dev/docs/md/quick-start.
Replace the placeholders in <angle brackets> before sending. Always review what the agent changes.
