Agent execution always follows the service account, tenant policy, and operator roles. Configuring a capability never grants access that the execution principal does not already have.
What the Agents workspace makes visible
The workspace provides Threads, Agents, Plugins, and Memory. In Agents, conversations appear in compact, expandable groups with each Agent’s avatar and name. Open a conversation to follow its messages and tool activity. Use the Agent menu to open Configure, Inbox, or Storage in the management panel, or open Observability for analytics and durable Runs. The Agent definition supplies the identity, instructions, principal, and trigger policy for these conversations.Agent definition
An Agent definition has three practical layers:
The service account is the runtime identity. The Agent can only read or change what that principal and the current platform policy permit. The owner controls visibility and management; ownership does not silently add runtime privileges.
Status
- Active accepts and executes new work.
- Paused temporarily stops new work while preserving the definition and history.
- Disabled retires or hard-stops the Agent until an operator changes its status.
Create an Agent
1
Define the responsibility
Give the Agent a narrow mission, clear success criteria, and explicit exclusions. Prefer one durable responsibility over a broad “do everything” description.
2
Choose the principal and owner
Select a service-account user with least privilege. Assign the Agent to the appropriate user or group owner.
3
Configure runtime behavior
Choose the model, reasoning effort, execution mode, planning policy, and persistent Agent instructions. GPT-6 Sol and GPT-6 Luna are the supported model choices; autonomous runs use Standard processing. Set concurrency and run limits for the expected workload. See the current model support table for the shared Assistant and Agent matrix.
4
Select capabilities
Enable the platform tools and attach only the plugins required by the mission. Choose Restricted, Auto Review, or Full Access as the default permission level, then add narrower capability-specific settings where needed.
5
Review the workspace
Use the managed workspace created with the Agent or link an existing authorized folder. The workspace root becomes the durable file boundary for runs.
6
Start with a bounded run
Submit one direct instruction, inspect the trace and outputs, then add inbox, schedule, or event delivery after the behavior is proven.
How work reaches an Agent
Choose New conversation for direct work in the inline message composer. Select an Agent by name inside the composer, then send your instructions. The conversation and composer stay visible while work is processing; you can edit a follow-up draft and send it after the current run stops.
Choose New Inbox message for queued work. It starts with Instructions; expand Schedule to choose a delay or date, or Advanced for priority, an optional Duplication key, and structured input. A title is optional: RevoEngine generates one from the task when omitted and keeps the Inbox label aligned with the conversation. The key prevents duplicate non-terminal Inbox work for the same Agent.
Choose a permission level
Use Auto Review for bounded responsibilities where the Agent should make progress without waiting on every routine action. Use Restricted when an operator must inspect every material effect. Reserve Full Access for trusted workflows with least-privilege identities, narrow capabilities, proven behavior, and an equivalent external control boundary.
See Approvals and plans for the detailed behavior of each level.
Runs and sessions
An Agent run moves through durable states:CREATED → PENDING → PROCESSING → FINISHED
A run can instead become ACTION_REQUIRED, ERROR, or CANCELLED. Capacity-limited or delayed work remains pending with an observable reason rather than disappearing.
A session links follow-up root runs into one durable continuation. Continuing a terminal run creates a successor; it does not rewrite the original run. This preserves history while allowing instructions such as “compare the result with last month.”
Operators can resume, retry, continue, or cancel a run. Ordered run events provide evidence across root, child, and handoff work.
Delegation
For suitable background or automation work, an Agent can split a plan into bounded child runs and merge their results. Configuration limits both the total number of child runs and the number that can run concurrently. A failure policy decides whether one child failure stops the parent or whether partial results may still be merged. Delegation is intentionally bounded. It is not an unrestricted process graph, and it does not expand a child Agent’s identity or permissions.Runtime boundary
Autonomous Agents use managed RevoEngine capabilities, attached plugins, Storage, platform resources, and explicitly selected terminals. Terminal selection is limited by the assigned service account and its current permissions; an Agent with no selected terminals has no terminal access. This is not an unrestricted host-shell grant. See Terminals.Operate from the Agents workspace
The conversation list uses the same compact navigation as Threads, grouped by Agent. Drag Agent groups to change your personal order; avatar status markers show whether an Agent is active, paused, or disabled. The compact Assistant provides Threads, Agents, and Plugins through the list menu, with search scoped to the selected conversation view. Agent conversations open in its shared tab strip. Each conversation has an options menu for Archive. Finish or cancel active work before archiving; archiving hides the conversation while retaining its durable Runs.- Configure opens the management panel for general, identity, runtime, tools, plugins, memory, and terminals. Agents inherit active instance default plugins, can attach additional plugins, and can select which attached plugins load at the start. The panel first opens at three-quarters width and remembers your resize. Save includes validation details when required fields are missing.
- Inbox opens queued work, its generated or supplied title, readable priority, availability time, and claim state.
- Storage opens the workspace file explorer directly.
- Observability → Agents provides Overview and Runs. Overview includes task outcomes, model and token usage, and workspace totals for storage usage, files, folders, and file types. Storage totals include accessible active and archived files, without counting folder sizes twice.
Production checklist
- Use a least-privilege service account.
- Keep the responsibility and success criteria measurable.
- Start with Restricted or Auto Review and explicit approval routing.
- Bound active root runs, child runs, concurrent children, and the execution budget.
- Add de-duplication keys for recurring or externally retried inbox deliveries.
- Require durable files only when the task genuinely needs a file deliverable.
- Inspect failed and blocked runs before enabling a wider trigger surface.
- Pause an Agent before changing behavior that could affect in-flight work.
Agent configuration
Configure identity, runtime, permissions, plugins, memory, workspace, and limits.
Operate Agents
Connect triggers, inspect runs, recover failures, and scale proven workloads.
Tools and skills
Understand the capabilities available to Assistant and Agent execution.
Memory and workspaces
Place durable knowledge and files in the correct store.