The status log agents hand-roll on every build: a live state (building / blocked / awaiting-human / done), a timestamped worklog, and a checklist — so a human can glance once and know whether their agent needs them. Built on the status component; awaiting-human surfaces the doc in your dashboard inbox.
A live status your agent keeps as it works: flip the state, append a line each step, tick the checklist. When it hits awaiting-human, this doc surfaces in your dashboard “needs me” inbox (and email, if enabled) — so you only look when you’re actually needed.
Anything the human should know — decisions made, open questions, links.
“Keep the status fence updated as you work: flip state (building → blocked / awaiting-human / done), append - <ISO-timestamp> what you did log lines, and tick the checklist. Set state: awaiting-human the moment you need a decision, credential, or review — that’s my signal to jump in.”