Skip to content
WebViewGold Vibe Coding to App Store
English
Get WebViewGold

Project memory workshop

A CLAUDE.md and AGENTS.md workshop: project memory that an app team keeps current

A memory file is the cheapest way to make a coding agent behave the same for everyone on the team, and the easiest thing to let rot. This focused session is about writing CLAUDE.md or AGENTS.md together for an app project and agreeing on how it stays true.

Free 15-minute call with our team, or call +49 761 519 751 50 (24/7, English + German)

  • Approval guarantee – or money back within 14 days
  • No coding needed
  • One-time payment, free updates
  • 24/7 human support
  • Made in Germany since 2014 · ISO 27001

Why a file this small gets its own session

Claude Code reads CLAUDE.md at the start of a session, Codex reads AGENTS.md. Both are plain Markdown, and both usually come from one person in a hurry. Then the build command changes, a second developer adds contradicting advice, and the agent follows instructions nobody remembers writing.

In an app project the file carries more weight than elsewhere. A web app that also ships as iOS and Android apps has rules an agent cannot guess: layouts must survive narrow screens, account deletion must not vanish in a refactoring, a new checkout can break a store rule. The session does not hand out a template, because two already exist: the CLAUDE.md template and the AGENTS.md template. It is about making such a file your own and keeping it alive.

What belongs in the file, and what does not

ContentIn the file?Reason
Commands for install, test, build and deployYesThe agent runs them in every task
The architecture in a few linesYesIt saves the agent from exploring the project again
Native app context and store rulesYesThey cannot be read from the code
Mistakes the agent has made repeatedlyYes, as short rulesThis is how the file learns
API keys, passwords, tokensNeverThe file is committed and read by tools
Instructions for one specific taskNoThey belong in the prompt or the ticket

How the session runs

  1. Collect the evidence

    Everyone brings a few recent cases in which an agent did the wrong thing. They are the raw material.

  2. Let the agent draft, then cut

    The agent proposes a file from the repository. The team deletes what is obvious from the code and keeps what an agent would otherwise get wrong.

  3. Add the app section

    The native context, the store rules that apply to your app and the features the wrapper provides, adapted from the template.

  4. Test it on a real task

    Run the same small task with the old and the new file and compare the results. A rule that changes nothing can go.

  5. Agree on one source

    If the team uses several agents, decide which file is the original and let the others point to it. The comparison of instruction files shows which tool reads what.

A prompt that audits the file you already have

Prompt for Claude Code or Codex, read-only
Audit the project memory file of this repository (CLAUDE.md or AGENTS.md). Do not change any files.

1. Check whether every command in the file still exists in the project.
2. List statements that contradict each other or the code.
3. List paths, tools or services the file mentions that are gone.
4. This web app also ships as native iOS and Android apps inside a WebView wrapper. Say whether the file explains that, and whether it covers mobile layout, the privacy page, account deletion and payments.

Report the findings as a table with the columns line, problem and suggested fix. Then propose a shorter version of the file in your answer.

Maintenance rules teams usually agree on

  • The file changes through pull requests, like code, and one named owner approves the change.
  • A mistake the agent makes twice becomes a rule; a rule nobody can explain gets deleted.
  • Whoever changes a command, a folder or a service updates the file in the same pull request.
  • The file stays short enough to read in a few minutes. Details live in linked documents.

Frequently asked questions

Do we need both CLAUDE.md and AGENTS.md?

Only if the team uses agents that read different files. In that case keep the content in one of them and let the other refer to it, so there is one place to maintain.

How long should the file be?

Short enough that every developer has read all of it. When it grows, move details into linked documents.

Can the agent write the file itself?

It can draft it from the repository, and that is a good start. The parts that matter most cannot be derived from the code: store rules, past mistakes and decisions the team has made.

Three ways to get your app into the stores

Same result, different amount of work on your side. All three come with the approval guarantee and 24/7 human support.

Do it yourself

$109 per platform, one-time

The WebViewGold license: a finished Xcode or Android Studio project. Paste your URL, set name and icon, switch on features, build and submit. Free updates forever.

See plans on the main page

Cloud Builder

$349 per license, one-time add-on

No Mac, no Xcode, no Android Studio: configure in the browser, we build both apps in the cloud and hand them to App Store Connect and the Play Console.

How the Cloud Builder works Try the free demo

Full Service

$698 per platform $798

Send us your URL and icon. Our team configures, builds and delivers your app within 72 hours, white-label, with source projects. Both platforms for $1,158.

Full Service details Powered by WeSetupYourWebViewApp

Winter Sale prices valid until December 31, 2026.

Talk it through with our team

Free 15-minute call: your app, your tools, your next step

Rather start right away? See plans and pricing or hand it over with Full Service.