# Vibe Coding Apps Script — reusable prompts

Replace bracketed details with your own fictional examples and verified project information. Never include secrets or unnecessary private data. These prompts work with your chosen coding assistant; the website does not call an AI model.

## Plan a first version
Build a small Apps Script prototype for [outcome]. Environment: [bound sheet / standalone / web app]. Input contract: [file selection, tab, exact headers, types]. Output: [observable result]. No-change boundary: [operations excluded]. Examples: [normal, empty, invalid, repeated]. First list assumptions, files, entry functions and service calls. Separate pure transformations from Google service adapters. Propose the smallest preview-only slice, setup steps and tests. Identify official references for methods that need verification.

## Make a focused change
Current behavior: [observed behavior]. Requested change: [one change]. Preserve: [existing rules and boundaries]. Relevant code: [small function]. Example: [input → expected output]. Propose a focused patch, explain changed behavior and list relevant regression cases. Flag any added permission, external request or side effect. Do not rewrite unrelated code.

## Starter challenge: clearer owner display
In the supplied task-audit starter, blank owners display as blank. Show `Unassigned` in the preview for an empty or whitespace-only owner. Keep source cells unchanged, preserve case-sensitive ID matching, status counts and row mapping. Explain whether the display rule belongs in the analyzer or the sidebar and why. Include a normal-owner and blank-owner regression case. Do not add writes or new Google services.

## Understand unfamiliar code
Trace [fictional record] through this function. Explain input types, outputs, branches and every Google service call. Separate assumptions from verified behavior. Identify any mutation, sending, sharing or external request. Do not edit code yet. Propose one fixture that could disprove your explanation.

## Repair a bug
Entry point: [function/action]. Small reproduction: [steps + fictional input]. Expected: [result]. Actual: [result + redacted error]. Relevant code: [function]. Already tried: [observations]. Identify the smallest plausible cause, give one diagnostic check and propose a minimal patch. Distinguish local logic tests from live Google integration checks. Preserve [invariants].

## Review proposed code
Review this Apps Script proposal against [requirements]. Identify unverified methods, input/type assumptions, range offsets, permission changes, side effects, repeated-run behavior and failure paths. Cite the official service reference for any API claim. Do not treat absence of obvious risky calls as proof of safety. Suggest focused changes and observable tests.

## Refactor without changing behavior
The verified behavior is [contract], with tests [cases]. Refactor [specific area] to separate pure logic from service adapters or remove duplicated rules. Keep public entry points and behavior unchanged unless explicitly discussed. Explain equivalence and provide a concise patch plus regression plan.

## Prepare a handoff
Document the actual current project: input contract, entry points, configuration, executing identity, authorized services, test evidence, limitations, owner and stop/recovery procedure. Distinguish implemented features from plans. Do not include credentials. For a versioned deployment, record the version and the endpoint actually tested.
