A UI for talking to agents
Creators do not need to open a terminal or inspect raw runs. They ask for work, review results, approve risky actions, and see what changed from one dashboard.
Creator Dashboard · Pro · agent UI
A dashboard where creators talk to agents.
Creator Dashboard is the UI and API layer between a creator and their agents. The creator works in a clear dashboard; agents work through the API, reporting status, creating tasks, attaching outputs, and asking for approval before anything publishes, sends, or replies.
Use · Build — then run it
The simple way to run your system
What the Dashboard really is
One interface for creator decisions, one API for agent work
The dashboard is not another automation canvas. It is the operating surface around your agents: a place where humans can ask, inspect, approve, pause, and correct work while agents use a stable API to update the same shared state.
An API agents can write to
Agents report status, create tasks, attach outputs, request approval, and mark runs complete through a server-side API instead of improvising their own interface.
A human control layer
The agent can prepare content, emails, research, and replies. The dashboard decides what is visible, what waits for approval, and what is allowed to ship.
The core loop
Creator input becomes agent work. Agent work becomes dashboard state. Dashboard state becomes the next creator decision.
Step 1
Creator asks
Start a run, approve a draft, edit a task, or change a schedule.
Step 2
Dashboard calls
The request is turned into a scoped API action for the selected agent or runtime.
Step 3
Agent works
The agent reads context, performs the workflow, and streams progress back.
Step 4
Dashboard updates
Tasks, approvals, library artifacts, status, and errors stay visible in the UI.
How it works
You talk through the UI. Agents talk through the API.
The Dashboard and the agent read and write the same shared state from two sides. You drive it by clicking — asking, reviewing, approving. The agent drives it over a small, scoped HTTP API — opening tasks, reporting progress, and asking for approval. Neither side talks to the other directly; the Dashboard is the surface in between.
You, in the UI
No terminal, no raw runs. Every decision is a click in the Dashboard.
- Ask for work, start a run, or change a schedule
- Review a draft, edit it, or send it back
- Approve what's allowed to publish, send, or reply
- Pause an app or correct a task mid-flight
The agent, on the API
The agent never improvises an interface — it calls back into the Dashboard.
- Open a task with its app, title, and payload
- Stream progress and update task status
- Attach finished outputs to the Library
- Request approval before any risky action ships
The endpoints an agent calls
- POST
/api/tasksAgent opens a task — app, title, payload - PATCH
/api/tasks/:idUpdate status: queued · running · needs_approval · done · failed - POST
/api/tasks/:id/approvalAgent requests a human approval gate - PATCH
/api/workflows/:id/statusMove an app's state: idle · running · waiting · error · paused - POST
/api/eventsAppend-only run log for the Today and app views
One bearer token per creator, keys server-side, single opt-in to the agent. Endpoint shapes are a design proposal — the preview below renders state shaped like this contract.
Inside the Dashboard
An interactive preview
Click through the sidebar to explore each surface. The approval queue sits at the center — nothing publishes, sends, or replies without your yes.
Creator Dashboard Hermes
Today
What ran, what's waiting on you, and what's next.
8
Apps installed
2
Running now
3
Waiting on you
14
Runs today
Waiting on you
3Reply to “When is part 2?”
ReplyYouTube Comments Monitor
Newsletter issue #42 outline
PublishNewsletter Research Desk
Sponsor email · Acme Health
SendSponsor Outreach CRM
Activity
Run started — scanning 40 channels
Trend Radar · now
6 new mentions tagged
Mentions & Reputation Watch · 3 min ago
Reply drafted — waiting for your approval
YouTube Comments Monitor · 12 min ago
Issue #42 outline ready for review
Newsletter Research Desk · 1 hr ago
Run failed — source link returned 404
Content Repurposing Factory · 1 hr ago
18 ideas ranked and delivered
Content Idea Finder · Mon 7:02am
How it's wired
A control plane, not a shared brain
The Dashboard is the operator layer on top of your agent runtime. It talks to your agent so your browser never has to — and your runtime keys never leave the server.
Your browser
The Dashboard you operate.
Creator Dashboard
Control plane — opens Tasks, moves status, holds the approval gate.
Your runtime
Hermes or OpenClaw — your isolated agent.
Your browser talks to the Dashboard — never directly to the agent.
Runtime keys and credentials stay server-side.
Your files, memory, and secrets stay isolated to you.
One screen across every app you run.
Every app reports in
Apps land on a Dashboard surface
Install an app from the hub and it runs on your agent — sending its work back to the right surface, and anything risky to the approval queue.
Run your apps in the Creator Dashboard.
Start free: install the apps you need, run them on your own agent, and approve what ships. When you outgrow it, Creator OS is the full control plane for teams.