Claude Skills by fujitoid

Advisory secret hygiene for chat, about-to-write, and about-to-run content. Use when the user pastes a secret, shares credentials in a message, or you are about to write a file or run a command that might contain an inline API key, token, or password. Offers `ka set` and prefers `ka run` for env injection. Same detection vocabulary as the secret_guard hook, but advisory — this skill has no enforcement power on its own.

One-shot workflow to migrate a project's existing plaintext secrets (.env files, hardcoded API keys, config files) into the key-amnesia vault. Use when the user asks to "migrate to key-amnesia", "move my secrets to the vault", or clean up plaintext credentials in a repo. Locates candidates by filename/pattern only, reports name + path (never values), and walks the user through `ka set` + code changes to `ka run`.

Use key-amnesia (`ka`) so agents can run commands with secrets without ever reading vault values. Check state first (`ka status` / `ka list`), then prefer `ka run --cwd DIR --secret NAME --as NAME=ENVVAR -- ...`. Never paste keys into chat, argv, or files the agent can read. Use when needing API keys, passwords, env injection, vault secrets, or `ka`/`key-amnesia` commands.