AltrinaAI Load secrets — VibeStudio activation (fallback)
VibeStudio keeps your API keys and secrets in one place and renders them to a
single env file. Terminals launched from VibeStudio source that file
automatically, so the secrets are normally already in your environment —
check first (e.g. [ -n "$OPENAI_API_KEY" ]). Reach for this skill only as a
last resort, when a command has actually failed because a key, token, or
environment variable is missing.
Load the secrets
Run this once, through eval, pointing at this skill's folder:
eval "$(bash ./activate.sh --print)"
(Use the absolute path to activate.sh if your shell isn't already in this
folder.) It exports every managed secret into the current shell and, where
possible, wires your shell startup files so shells started later inherit them
too. It prints only the variable names it activated — never the values.
Sandboxed agents (read-only HOME, fresh shell per command)
Some agents (for example Codex) run each command in a fresh shell with a
read-only HOME, so a separate activate step doesn't persist and the
startup files can't be patched. There, source the env file in the same
command that needs the secrets — this only reads a file, so it works even
when HOME is read-only:
. "${VIBESTUDIO_ENV:-$HOME/.config/vibestudio/env}" && your-command
If it reports that no secrets are configured, add them in VibeStudio and run it again.