Cloud Agent and Automations

The Cloud Agent helps explore and maintain a Cloud Brain. Its conversations belong to a Workspace and can be renamed, archived, deleted, or cancelled from the Cloud app.

Choose the Agent scope

Start by choosing either Workspace-wide read access or a specific Project. Workspace-wide work can inspect accessible Brains; a selected Project gives the Agent the context it needs to make Brain changes. Review the selected Project before asking the Agent to write knowledge.

The Agent can list, read, search, create, update, archive, and delete Pages and root pages within its permitted scope. It can also use web research when available, subject to per-response limits. Treat its output as proposed project knowledge: review the resulting Page, timeline, and revision history before relying on it.

Create recurring automations

You can create an automation from a natural-language goal, continue the design in an Agent conversation, then set its instructions and schedule. Use one for a repeatable review such as checking a Project for stale assumptions or preparing a regular knowledge summary.

Before enabling one, review the proposed instruction and scope. You can pause, resume, run, retry a failed run, or delete a custom automation, and inspect its run history and proposals. Schedules are shown in your local time and saved as UTC; do not manually convert them. Do not assume an automation silently applies changes—review its output before accepting knowledge changes.

Brain health checks

Each Workspace has a system-managed Brain Health Check. Owners and admins can pause or resume it, choose its daily schedule in their local time, run it manually, or retry a failed run. Members can inspect runs and findings.

The check reports issues rather than rewriting content. It validates Page identifiers, category, status, tags, dates, timeline format, fixed root-page identity, and missing wiki-link targets within a Brain. Run details and findings are retained for a limited period, so resolve or record important findings promptly.

Health checks are system-managed shared quality guardrails. They are separate from the recurring automations your team creates.