Understand projects, services, and execution
A project contains code and configuration that can call supported services under an authorization context. A script attached to a document has different packaging from a standalone project. Start by identifying the account, files, and external interfaces involved. This page owns platform concepts; the project guide walks through a small review-queue example.
- ✓ Use a test project and non-sensitive sample records.
- ✓ Consult current official documentation for supported services and setup.
Treat permissions and triggers as design decisions
Authorization determines what the code can access, while triggers determine when and under whose context it runs. Review requested scopes and durable ownership before scheduling. Keep sensitive values out of public source and sheet cells. Execution time and service quotas constrain workload size; their current values should be checked rather than copied indefinitely into a guide.
- ✓ Request the least access needed for the task.
- ✓ Document who owns triggers and how to disable them.
Learn reliable behavior before adding features
Handle empty inputs, invalid rows, failed requests, and repeated runs explicitly. Write clear logs without exposing personal data or secrets. Prefer small functions with predictable inputs and outputs, then verify expected results on a known dataset. A manual run should be understandable before the script becomes an unattended operational dependency.
- ✓ Keep an original-data backup and recovery instructions.
- ✓ Move to the project example when the core concepts are clear.