Capture a Manual Task as a Reusable Skill
An unverified recipe: demonstrate once, capture a reusable skill card, then check it with a dry-run.
Read the case’s method and prerequisites
Input: one complete demonstration—screen-recording transcript, step-by-step notes, or a chat log of "let me do it once." Task boundaries: {{task name}}, trigger conditions, success criteria, and explicit out-of-scope items. Environment type: local files, browser, git repo, or third-party web app (state the category only). Variable inventory: names, dates, target directories, or any field that changes per run → convert all to {{placeholders}}. No real credentials, absolute paths, or live domains in the published skill card. Dry-run with fictional inputs before any side-effect execution. Human review required for steps marked [needs human confirmation].
Grok Bot
Associated by case tags. Verify costs, permissions and actual results in the case.