💻 Coding Prompt
Gemini for Mobile Devs: Write a README for a New Codebase
Intermediate Gemini prompts for Mobile Developers writing READMEs that reduce onboarding time for new team members
The Prompt
You are a senior mobile engineering documentation specialist with 9 years of experience writing technical documentation for iOS and Android codebases at digital agencies where a well-structured README is the difference between a new developer being productive in two days or two weeks on a client project handoff. Help me write a README so I can cut code review time in half and produce a codebase README that gives a new mobile developer everything they need to set up the project, understand the architecture decisions, and contribute code that passes review without requiring constant senior developer intervention.
My situation:
- Project type and platform: [e.g., "a React Native application for a retail client — the app is a customer loyalty and rewards platform with iOS and Android builds, approximately 28,000 lines of code across 140 components"]
- Project setup complexity: [e.g., "the project requires Node 18.x, specific versions of CocoaPods and the Android SDK, three environment files (.env.development, .env.staging, .env.production) that are not in the repository, and a manual step to link a native module that the auto-linker does not handle"]
- Architecture decisions that confuse new developers: [e.g., "the state management uses a hybrid of Zustand for UI state and React Query for server state — previous developers have repeatedly asked why Redux was not used, and the code review comments show repeated attempts to add Redux slices — the decision is not documented anywhere"]
- Code review failure patterns: [e.g., "the three most common code review rejection reasons in the last 6 months are: components written without the project's custom hook pattern, API calls made directly in components rather than through the React Query layer, and styles written with StyleSheet.create in new files rather than the existing theme system"]
- Team and handoff context: [e.g., "the project is being handed from the agency's build team to a client-side development team of four developers who have React Native experience but have never seen this codebase — the handoff is in 3 weeks"]
- Documentation currently available: [e.g., "a Confluence page with the project kick-off brief from 18 months ago and three ADR documents that are partially complete — no setup guide, no architecture overview, and no contribution guidelines"]
- Review time target: [e.g., "the current average code review cycle is 4.2 round trips per PR — the target is 2.1 round trips, which requires the README to preemptively answer the questions that cause the most review comments"]
Deliver:
1. A complete README.md document — organized in seven sections: project overview (purpose, platform, client context), prerequisites (exact version requirements for Node, CocoaPods, Android SDK, and the three environment files with placeholder values and field explanations), installation steps (numbered commands from clone to first successful build on both iOS and Android), architecture overview (the Zustand-plus-React-Query hybrid with the rationale for not using Redux written as a permanent decision record), the three mandatory contribution patterns (custom hook pattern, React Query layer requirement, and theme system usage), the code review checklist the submitting developer must complete before opening a PR, and the escalation path for questions during the handoff period
2. A setup troubleshooting guide — a dedicated section covering the seven most common setup failure points for this project (the native module manual link step, the environment file sourcing, the CocoaPods cache issue on Apple Silicon, and four others the build team has documented verbally), each with the error message the developer will see, the cause in one sentence, and the exact command or file change that resolves it
3. An architecture decision record for the state management choice — a complete ADR document in the standard format (title, status, context, decision, consequences) that explains the Zustand-plus-React-Query decision, the alternatives considered, and the consequences — designed to be linked from the README and to permanently close the Redux discussion in code review
4. A contribution pattern guide with code examples — three annotated code examples showing the correct implementation of the custom hook pattern, the React Query data fetching pattern, and the theme system styling pattern — each example paired with an incorrect version showing the pattern the code review commonly rejects, and a one-sentence explanation of why the correct pattern is required
5. A pre-PR checklist — a markdown checklist the developer includes in every PR description, covering the eight criteria the senior reviewer checks first (correct hook pattern used, no direct API calls in components, theme system used for all styles, no new dependencies added without a comment, unit tests added for business logic, no hardcoded strings, accessibility props on interactive elements, and no console.log statements in production code)
6. A onboarding sequence brief — a recommended 5-day onboarding plan for the four incoming developers, covering day 1 (README setup and first build), day 2 (architecture walkthrough and ADR review), day 3 (first small feature ticket using the contribution patterns), day 4 (code review simulation where the senior developer reviews the day 3 output against the checklist), and day 5 (first independent PR submission)
7. A README maintenance protocol — a set of rules for keeping the README accurate as the codebase evolves, covering which sections require an update when a dependency version changes, when a new architecture decision is made, or when a new code review rejection pattern emerges — and the GitHub Actions check that fails if the README has not been updated in more than 90 days
**Write the README and every supporting document as if it will be read by a React Native developer on their first morning on the project, with their laptop open and no senior developer available to answer questions — every step must be executable without prior context, every architecture explanation must answer the question a developer will ask before they ask it, and every code example must be copy-paste functional rather than illustrative.**
💡 How to use this prompt
- Write the setup troubleshooting guide from output item 2 before the main README sections. The most expensive part of a codebase handoff is the first-day setup failure — a developer who cannot build the project on day one loses confidence and consumes senior developer time. The troubleshooting guide for the native module manual link step and the environment file sourcing should be written first because these are the two setup failures that generate the most Slack messages from new developers in the first 48 hours.
- The most common mistake is writing the architecture overview section with accurate technical content but without the rationale for the decisions. A new developer who reads "this project uses Zustand for UI state and React Query for server state" and does not understand why will continue to open PRs with Redux slices until the review friction trains them out of it. The architecture overview must lead with the decision rationale, not the implementation description.
- Gemini's real-time web access is useful for pulling the current recommended React Native setup commands, CocoaPods version compatibility notes, and Android SDK setup documentation before writing the installation steps. For the architecture decision record, the contribution pattern guide, and the pre-PR checklist, paste Gemini's research into Claude for cleaner, more precise technical documentation language.
Best Tools for This Prompt
🤖 Best AI Coding Tools for This Prompt
Tested & reviewed — run this prompt with the best AI tools
Related Topics
About This Coding AI Prompt
This free Coding prompt is designed for Gemini and works with any modern AI assistant including ChatGPT, Claude, Gemini, and more. Simply copy the prompt above, paste it into your preferred AI tool, and customize the bracketed sections to fit your specific needs.
Coding prompts like this one help you get better, more consistent results from AI tools. Instead of starting from scratch every time, you can use this tested prompt as a foundation and adapt it to your workflow. Browse more Coding prompts →