Activity Feed
Think of the Activity Feed as a security-camera roll for your workspace: whenever your connected assistant does something in Nirvai, it's recorded — what it did, which assistant did it, when, and a door straight to the result. Nothing important is trapped in a chat you'll scroll past and lose.
Why it exists
When an AI assistant acts on your behalf, you want to see it — not take it on faith. Because your assistant does its thinking on its side and only the actions run in Nirvai, the feed is the honest ledger of those actions. It's grouped into runs — one run per task — so a whole "build me a sales dashboard" job reads top to bottom.
- LIVE APPProposed live applive_apps.propose08:11 PM
- LIVE APPTested the datalive_apps.test_script08:12 PM
- LIVE APPBuilt live applive_apps.stage08:13 PM
What each row tells you
- Which assistant — the provider mark and name (Claude Code, Codex, …), so you always know who did what.
- What it did — a plain-language label, with the exact action name beside it for the technically curious.
- What it touched — a tag for the kind of thing (a live app, a credential, a database…) and a link straight to it in Nirvai.
- When — every action is timestamped; a run shows its start and end.
The feed keeps the inputs and links to results — never the raw output of an action. Big results live where they belong (a live app, a file), and the feed just points at them.
Open a run for the full picture
Click any run to open its details — the complete timeline plus everything the task produced: the setups it proposed (each with its own review page) and the files it created, browsable and downloadable. It's the one place that answers "what happened in this session, and where did it all go?"