Mock AGUI Streaming Server

This is a lightweight mock backend used to simulate AGUI streaming behavior while the real backend is under development.

The goal of this server is to:

• Enable UI QA testing
• Simulate realistic streaming responses
• Mimic backend success and failure scenarios
• Keep the implementation simple and easy to modify

---

Overview

The server exposes two endpoints:

• POST /orgs/:orgId/agents/:agentId/answer
• POST /orgs/:orgId/agents/:agentId/follow-up

Both endpoints stream events using Server-Sent Events (SSE) following the AGUI protocol.

The responses are hard-coded but dynamically adapted based on the input question.

---

Dynamic Behavior

The mock server supports simple keyword-based behavior to simulate real backend conditions.

1. Simulated Errors

If the question (q in request body) contains the word:

error

The server will:

• Emit RUN_STARTED
• Emit header
• Emit RUN_ERROR
• End the stream

This allows QA to validate error states and UI error handling.

---

2. Disabling Follow-Up (Answer Endpoint Only)

If the question contains the word:

disabled

The /answer endpoint will send:

{
  "followUpEnabled": false
}

inside the header custom event.

Otherwise, followUpEnabled defaults to true.

This allows QA to verify UI behavior when follow-up functionality is disabled.

---

3. Realistic Answer Simulation

The generated response:

• Is hard-coded
• Streams content in small chunks
• Simulates step progression (searching, thinking)

This makes the UI behave as if it is connected to a real AI backend.

---

Endpoint Details

1. /answer

Purpose

Generates the initial (head) answer.

Behavior

• Creates a new threadId

• Sends:

  • RUN_STARTED
  • header (with followUpEnabled)
  • STEP_STARTED / STEP_FINISHED (searching, thinking)
  • TEXT_MESSAGE_CHUNK (streamed answer)
  • RUN_FINISHED

---

2. /follow-up

Purpose

Generates a follow-up answer in an existing conversation.

Behavior

• Uses conversationId from request body as threadId

• Sends:

  • RUN_STARTED
  • header
  • STEP_STARTED / STEP_FINISHED (thinking)
  • TEXT_MESSAGE_CHUNK
  • RUN_FINISHED

This endpoint does not control followUpEnabled.

How to Start the Server

1. Install dependencies

npm install

2. Start the server

node server.js

3. Server runs at

http://localhost:3000