AI manage environment secrets

Updated my AI skill for deploying apps to your own server.

kamal-deploy-skill 1.8.0: secrets are now per-environment 🔐

Each environment gets its own file:
🟢 bin/secrets production edit
🟢 bin/secrets staging edit

https://github.com/kalashnikovisme/kamal-deploy-skill/

Comments

[Alex Shev]

AI can help inventory and rotate secret references, but it should not blur the boundary between planning and secret access. A safe design keeps the model on metadata and policy, routes actual retrieval through short-lived scoped credentials, and logs each action for review. The important control is proving that the assistant could not exfiltrate a value while still being useful.