Chat

Ask questions and get answers with citations. See Answer mode.

Method	Path
POST	/api/v1/chat

Chat

POST /api/v1/chat

Answer a question over a case with citations and RAPTOR clusters. With stream: true the response is a Server-Sent Events stream (delta, citations, done, error); see Chat & streaming.

From the API schema: Ask a question to the corpus’ RAGFlow assistant.

With stream: false (default) returns a ChatResponse. With stream: true the response is text/event-stream with these events:

• delta — \{"text": "..."\} incremental answer text
• citations — \{"citations": Citation[], "clusters": AnswerCluster[]\} once the answer is complete
• done — \{"sessionId": "..."\} (pass it back as sessionId to continue the conversation)
• error — \{"message": "..."\}

Citations point to graph nodes (nodeIndex), so they can be focused in the 3D viewer. clusters are RAPTOR thematic summaries with the node indices they cover.

Auth: API key (X-API-Key) or user session

Request body

{
  "case": "papers",
  "question": "Which methods combine knowledge graphs with LLMs?"
}

cURL

curl -X POST "https://vol.example.com/api/v1/chat" \
  -H "X-API-Key: $VOL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"case":"papers","question":"Which methods combine knowledge graphs with LLMs?"}'

TypeScript

import { VolClient } from "@vol/sdk";

const vol = new VolClient({ baseUrl: "https://vol.example.com/api", apiKey: process.env.VOL_API_KEY });
await vol.chat({ case: "papers", question: "Which methods combine knowledge graphs with LLMs?" });

// streaming
for await (const ev of vol.chatStream({ case: "papers", question: "…" })) console.log(ev);

Python

import os, requests

r = requests.post(
    "https://vol.example.com/api/v1/chat",
    headers={"X-API-Key": os.environ["VOL_API_KEY"]},
    json={
      "case": "papers",
      "question": "Which methods combine knowledge graphs with LLMs?"
    },
    timeout=60,
)
r.raise_for_status()
print(r.json())

Response 200

{
  "answer": "Several works ground LLMs on citation graphs [1] and summarise evidence recursively [2].",
  "sessionId": "c6f1d2…",
  "citations": [
    {
      "nodeIndex": 412,
      "nodeId": "paper_412",
      "title": "Graph-augmented retrieval for scientific QA",
      "snippet": "We combine a citation graph with dense retrieval…",
      "score": 0.87,
      "docName": "paper_412.pdf",
      "chunkId": "8c1e…"
    }
  ],
  "clusters": [
    {
      "id": "c_17",
      "title": "Graph-based retrieval",
      "summary": "Methods that use knowledge graphs to ground LLM answers.",
      "nodeIndices": [
        412,
        977,
        1503
      ]
    }
  ]
}

Types

interface ChatRequest { case: CaseId; question: string; sessionId?: string; stream?: boolean }

type CaseId = "papers" | "cordis" | "patents" | "trials";

interface ChatResponse { answer: string; sessionId: string; citations: Citation[]; clusters: AnswerCluster[] }

interface Citation {
  nodeIndex: number;
  nodeId: string;
  title: string;
  snippet: string;
  score: number;
  docName?: string;
  chunkId?: string;
}

interface AnswerCluster { id: string; title: string; summary: string; nodeIndices: number[] }