Back up your custom AI coding skills, automation rules, and project files to cloud storage without accidentally including passwords or API keys.
Restore a working AI coding workbench on a new machine from a cloud backup without having to reconstruct every setting from memory.
Keep old timestamped backup archives from piling up in cloud storage using the built-in archive naming and cleanup rules.
Fill in a configuration file with your local paths and chosen cloud provider before running the scripts.
This project is a backup skill for Codex, an AI coding assistant. Its purpose is to create a recovery package for a local AI workbench, meaning the collection of custom scripts, automation rules, project files, and personal instructions that someone builds up over time while working with AI tools. The problem it addresses is that a standard file backup tends to capture the files but miss the context: the instructions, runbooks, and preferences that explain how those files fit together and how to restore a working setup from scratch. A full disk backup captures that context, but it also captures sensitive material like passwords, API keys, browser sessions, and wallet credentials that should never end up in a cloud archive. This skill sits between those two approaches. It builds a restore package that includes custom skills and automations, project source files, sanitized environment templates, and a distilled record of working preferences, while deliberately excluding credentials and private account data. The backup skill itself is included in the package so that the most current backup rules are always preserved alongside the files. The typical workflow involves filling in a configuration file with your local paths, cloud provider, and project names, then running three PowerShell scripts in order: one that builds the package, one that verifies it, and one that handles the confirmed upload cycle. The tool supports a range of cloud storage destinations including Baidu Netdisk, OneDrive, Google Drive, Dropbox, and iCloud Drive, all configurable through the same config file. It also manages archive naming so that old timestamped backup folders do not pile up in your cloud storage. The project explicitly does not store passwords, keys, cookies, browser profiles, or any account secrets. Those still require a separate encrypted vault or manual login after a restore.
← cbl980712-coder on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.