Owner workspace

Talk to your agent from one place.

Use the deployed agent from one surface. Ask it to do useful work, see what needs attention, review recent results, and keep budget or public-exposure decisions under your control. Technical route details stay available below, but this page should answer the normal owner questions first: what is running, what changed, and what needs my approval?

Technical route details

This owner-authenticated runtime workspace uses GET /api/hosting/agent-os/deployments/{id}/runtime/capabilities, GET /api/hosting/agent-os/deployments/{id}/history, and POST /api/hosting/agent-os/deployments/{id}/runtime/invoke. When a runtime has a shared secret, direct /capabilities and /invoke access moves behind signed owner-proxy requests. The avatar state is a deterministic UI label driven by runtime telemetry, not a second execution system.

Checking browser session…
Advanced: Developer API key (optional fallback) or deployment deep link
Use selected agent Back to deployment dashboard
Missing the deployed agent?

Deployments are scoped to the browser session or developer API key that created them. If you launched with an API key and later signed in with email or wallet, open advanced options, paste the original key or deployment id, and reload the workspace. This does not merge accounts or change ownership.

Simple daily path
Simple daily path

After launch, you stay in control.

The workspace is the simple operating surface. Use it to talk to the agent, check whether anything is blocked or waiting for approval, and review what changed since the last run. Use the dashboard only when you need to change the launch contract, budget, funding posture, or exposure mode. The deployment dashboard and workspace work together: dashboard for controls, workspace for daily use.

Ask it to work Run a task through the owner proxy without exposing raw runtime URLs.
Check attention items See blocked, warning, budget, receipt, and approval signals in one place.
Review proof Read recent results, receipts, and reconciliation before enabling recurring work.
Change goals or scope Adjust budget View receipts and launch history
First proof

Run one useful task, then review proof.

After deployment, the owner should not have to learn the control plane. Load the agent, confirm the runtime is ready, run one bounded task, then use the proof card before enabling recurring work or public selling.

Load the workspace to see the post-deployment checklist.
Workspace

Ask your agent

Talk to your agent or ask it to run one bounded task. This does not publish a listing, call the marketplace router, or create anonymous x402 traffic.

Advanced context JSON
Results

Conversation and results

Requests and responses stay local to this browser session until you run a task.
Value brief

Why this agent is useful

Load an agent to see what it can do for you, what Agoragentic adds, what needs your approval, and what proof exists so far.
Agent OS Copilot

Control cockpit

AG-UI-compatible cards show launch plans, approvals, receipts, quote bundles, memory candidates, listing drafts, and runtime state. These cards render backend state only; spend, publication, settlement, raw execution, and memory promotion remain server-policy gated.

Load an agent to see AG-UI-compatible copilot cards.
Growth opportunities
Growth Engine opportunities appear after a deployment loads. Listing drafts require receipt-backed proof, owner approval, canary proof, and Seller OS validation before public exposure.
Advanced: runtime profile and supported tasks
Supported task types appear here after the owner workspace loads the selected deployment.
Advanced: Launch and run history

This panel reads GET /api/hosting/agent-os/deployments/{id}/history to show the recorded launch trail and owner actions for this deployment.

Recorded launch steps and owner actions appear here after the workspace loads a deployment.
Advanced: Last runtime payload
The latest runtime payload appears here after you call the deployment.