The idea is about sharing secrets and keys not locking people out of the codebase.
Let's say you are using an OpenAI api key. The only people who can use your key are the ones you added their public key to the vault.
If you decide they can no longer have access, you change the secret and pull their public key from the repo.
This is no different then if you were sharing secrets in a .env except that they are not stored in plain text which is why you can upload it to github.
If you never give anyone access to the vault they can never use your secrets even if they have the code. They can create their own vaults and keys but they can't use yours.
Was thinking about how to make leaking keys less of an issue, especially as ai loves to just commit stuff. And why not store them in memes.
Memevault is a zero-dependency, single-binary secret manager that lets you keep encrypted project secrets anywhere by storing them inside a meme image. Instead of sharing `.env` files or relying on ad hoc workflows, teams can grant and revoke access per user, and the vault is re-encrypted accordingly. It works consistently across Windows, Linux, and macOS, injects secrets only into the environment of the command you run (`memevault run -- ...`), and includes key rotation plus a `scan` command to catch secrets referenced in code but missing from the vault.
Let's say you are using an OpenAI api key. The only people who can use your key are the ones you added their public key to the vault.
If you decide they can no longer have access, you change the secret and pull their public key from the repo.
This is no different then if you were sharing secrets in a .env except that they are not stored in plain text which is why you can upload it to github.
If you never give anyone access to the vault they can never use your secrets even if they have the code. They can create their own vaults and keys but they can't use yours.