Frequently asked questions
What's the difference between my personal project and a team project?
Your personal project is visible only to you — a sandbox for experimenting. A team project shares its agents, pipelines, toolkits, and conversations with every member, according to their role. Switch between them with the project selector at the top of the sidebar.
Do I need a credential for every agent?
Only if the agent uses a toolkit that requires one. A plain agent with no toolkits needs only a model — configured once for the whole project in AI Providers.
Can I get my agent's exact instructions back after I lose them?
Yes, as long as they were saved as a version. Use the version dropdown on the agent to switch to an earlier version, or export a version as a backup file.
Why does my agent stop mid-response?
Its model likely hit the configured Max Completion Tokens limit. Increase it in the agent’s model settings — and check any nested agents’ limits too. See Troubleshooting.
Can I use the same credential across multiple toolkits?
Yes — one credential can back several toolkits and agents. See Credentials.
Why is a toolkit type missing from the create list?
Which toolkit types are available depends on the worker your deployment runs, and on any restriction your administrator applied. See the toolkit catalogue.
Is there a way to recover a deleted secret?
No — deleting a secret is permanent, and every credential referencing it fails authentication immediately. Consider hiding a secret first if you only want to stop new usage while confirming nothing still depends on it.