Streaming Responses

SDAC Chat Streaming

POST /sdac/chat

The source-backed streaming endpoint for SDAC report review. Browser widget code normally calls the dashboard/widget proxy at POST /api/ingestion/sdac/chat, which forwards to the runtime while preserving the widget token boundary. APIM may publish the runtime route with an environment-specific prefix.

curl -X POST "$BASE_URL/sdac/chat" \
  -H "Content-Type: application/json" \
  -d '{
    "reportId": "8201EDC2-2EDE-4CA1-AF44-D0F5AA185CDB",
    "message": "What is the fringe variance for this report?",
    "userId": "auditor@state.gov",
    "sessionId": "browser-session-abc123",
    "conversationId": null
  }'

Request fields:

Field	Type	Required	Description
reportId	UUID	Yes	Cost report identifier
message	string	Yes	User's question or instruction
userId	string	Yes	Auditor email or identifier
sessionId	string	Yes	Browser session identifier
conversationId	UUID	No	Omit for new conversation; include to continue

SSE Event Stream

The response is a Server-Sent Events stream with these event types:

metadata -- Sent first, contains session context:

{
  "conversationId": "conv-uuid-123",
  "turnNumber": 1,
  "isNewConversation": true,
  "conversationExpired": false
}

tool-start -- Agent is calling a validation tool:

{
  "toolCallId": "call-uuid",
  "toolName": "sdac-validate-fringe",
  "displayName": "Validate Fringe Benefits"
}

tool-result -- Tool execution completed:

{
  "toolCallId": "call-uuid",
  "toolName": "sdac-validate-fringe",
  "success": true,
  "summary": "3 warnings found in fringe data"
}

delta -- Incremental text from the agent:

{ "content": "Fringe is " }

usage -- Token consumption for cost tracking:

{
  "promptTokens": 1200,
  "completionTokens": 85
}

done -- Stream complete:

{ "success": true }

error -- Error during processing:

{ "message": "Tool execution failed" }

Client-Side Parsing

const response = await fetch('/api/ingestion/sdac/chat', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    reportId: '8201EDC2-2EDE-4CA1-AF44-D0F5AA185CDB',
    message: 'What is the fringe variance?',
    userId: 'auditor@state.gov',
    sessionId: 'browser-session-abc123'
  })
});

const reader = response.body.getReader();
const decoder = new TextDecoder();

while (true) {
  const { done, value } = await reader.read();
  if (done) break;

  const chunk = decoder.decode(value);
  // Parse SSE events from chunk
  for (const line of chunk.split('\n')) {
    if (line.startsWith('data: ')) {
      const data = JSON.parse(line.slice(6));
      // Handle event data
    }
  }
}

---

Agent Streaming (Generic)

POST /ai/{domain}/agents/{slug}/stream

Any agent can be invoked in streaming mode. See Invoke Agents & Tools for input format details.

Example -- TAP Image Caption Agent:

curl -X POST "$BASE_URL/ai/tap/agents/image-caption/stream" \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [
      {
        "role": "user",
        "content": [
          { "type": "image", "image": "https://example.com/image.jpg" },
          { "type": "text", "text": "{\"context\": {\"postId\": 12345}}" }
        ]
      }
    ]
  }'

The SSE event types are the same as described above (delta, tool-call, tool-result, usage, done, error).