Playground

The Playground is an interactive testing panel in the canvas editor. It sends real HTTP requests to your deployed workflow endpoint — no external tools needed.

The workflow must be deployed before using the Playground. The Playground tests the live endpoint, not the canvas draft. Use the canvas Test button for pre-deployment testing.

Click the Playground tab in the bottom bar of the canvas editor. It appears alongside Test History, Deployments, Variables, and other tabs.

The Playground automatically detects the right testing mode based on your workflow's trigger:

• API Mode: When the API trigger's protocol is rest (default) or when using a Webhook trigger
• Chat Mode: When the API trigger's protocol is openai-chat

API Mode provides a Postman-style interface for testing REST endpoints.

Features:

• Method selector: Choose GET, POST, PUT, PATCH, or DELETE
• JSON body editor: Write request bodies with syntax highlighting
• Response viewer: See status code, headers, and response body
• Timing: Response time displayed for each request

How to use:

1. Select the HTTP method
2. Enter a JSON body (for POST/PUT/PATCH)
3. Click Send
4. View the response in the panel below

Chat Mode provides a conversational interface for AI agents. It's activated automatically when the API trigger uses the openai-chat protocol.

Features:

• SSE streaming: Tokens appear in real-time as the agent generates a response
• Session continuity: The Playground sends X-Tensorify-Session-Id automatically, so multi-turn conversations work with memory plugins
• Message history: See the full conversation in the panel
• Tool call visualization: When the agent calls tools, the tool name and result are shown inline

How to use:

1. Type a message in the input field
2. Press Enter or click Send
3. Watch the agent's response stream in real-time
4. Continue the conversation — memory persists across messages

To use Chat Mode, your workflow needs:

1. An API Endpoint trigger with protocol: openai-chat
2. An AI Agent node with streaming: true
3. A sessionKey on the agent or memory plugin bound to {{ api_request.headers["x-tensorify-session-id"] }}

Feature	Playground	Canvas Test	Run Button
Target	Deployed endpoint	Canvas draft	Deployed (manual trigger)
Requires deployment	Yes	No	Yes
Streaming	Yes (Chat mode)	No	No
Real HTTP request	Yes	No (simulated)	Yes (API call)
Auth tested	Yes	No	No

• API Endpoint Reference — trigger configuration including OpenAI Chat protocol
• AI Agent Reference — streaming and tool settings
• Build an AI Agent — end-to-end guide