What is the ClawLink Playground?

The Playground is a chat page in the ClawLink dashboard where you type a request in plain language and ClawLink runs it against the apps you have connected. It exists so you can check that a connection works, and see what an app can do, before you point your own agent at it. It is not the product: your day-to-day work still happens in OpenClaw, Hermes, the CLI, or an MCP client.

How to use it

1. Connect at least one app in Dashboard → Integrations. The Playground has nothing to run until you do.
2. Open Dashboard → Playground.
3. Type a request, such as "show my last five emails" or "list my Notion databases", and send it.
4. Reads run right away. Writes (send an email, create a page, post a message) pause and show you an approval card with the exact values that will be used. Nothing is sent until you press Approve.
5. To focus on one app, start the message with @ and pick it, for example @gmail find the invoice from Acme. Without a mention, ClawLink picks from all your connected apps.

Chats are saved in the sidebar under Playground. The pencil icon starts a new one.

Is the Playground the same as my agent?

No. Your agent (OpenClaw, Hermes, Claude Code, or any MCP client) is separate software that you install and run. ClawLink gives it access to your apps through a pairing step and an API key. The Playground is a small agent that ClawLink runs for you inside the dashboard, with the same connections.

What is the same:

• The connected apps. An app you connect once is available to both.
• The tools. The Playground runs the same actions your agent would, through the same hosted runtime, so a request that works here will work there.
• The rules. Write actions ask for approval in both places, and both count against the same tool-call quota.

What is different:

• The Playground only exists inside the ClawLink dashboard. It does not see your files, your terminal, or anything on your computer.
• Your agent has its own model, memory, and instructions. The Playground has none of those; it is a plain chat with a short memory of the current conversation.
• The Playground has its own request allowance (below). Your agent's tool calls are limited only by your plan's tool-call cap.

If something works in the Playground but not from your agent, the connection is fine and the problem is on the agent side: pairing, an old plugin version, or the agent choosing the wrong tool. If it fails in both, reconnect the app.

What it is good for

• Checking a new connection. Ask for a simple read (your profile, a list of recent items). If it returns data, the connection is healthy.
• Finding out what an app can do. Ask "what can you do with my Trello?" and it lists the available actions.
• Trying a write safely. The approval card shows exactly what will be sent, so you can confirm the shape of a request before automating it.
• Running a one-off task without opening your agent.

Limits

Each message you send is one Playground request. The allowance depends on your plan:

Approving a write spends a second request, because the conversation is sent again to finish the action. Any app action the Playground runs also counts as a tool call against your plan's normal tool-call cap, exactly as it would from your agent.

The Playground needs an active trial or subscription. Long transcripts are trimmed from the top so the newest messages always fit.

Privacy

The Playground shows only the results of the actions it ran and the prose answer. It never displays connection ids, credentials, or raw tool schemas. Approval cards show the arguments of a write, and that is deliberate: what you see is what runs.

---

ClawLink connects 1,000+ apps to agents like OpenClaw and Hermes: browser login, hosted OAuth, no API keys. Start free or browse integrations.