AI Assistants

Chat with Document

Chat with a document using conversational question-answering.#

POST/lmkit/v1/chat-with-document

Loads a document and enables multi-turn conversational Q&A powered by RAG.

Session lifecycle:

First call: provide 'input' + 'input_format' to create a session. Returns a 'session_id'.
Subsequent calls: provide 'session_id' + 'question' to chat. Conversation context is maintained.
DELETE /lmkit/v1/chat-with-document/ to release resources.

Response modes:

• stream: false (default): returns a single ChatWithDocumentResponse JSON object.

• stream: true: returns a text/event-stream of SSE events. Each event is data: {json}\n\n containing a ChatWithDocumentStreamChunk object with these fields:

Field	Type	Description
`session_id`	string	The session identifier
`delta`	string	Incremental text token (empty in the final chunk)
`thinking`	bool	`true` if the delta is internal reasoning (chain-of-thought), `false` for the user-visible answer
`done`	bool	`true` on the final chunk
`response`	string?	Complete answer text (final chunk only, excludes thinking)
`source_references`	array?	Retrieved document passages (final chunk only)

The stream ends with data: [DONE]\n\n.

If document loading exceeds the configured timeout, returns 202 Accepted with a job_id. Poll GET /lmkit/v1/jobs/ for status and results.

Request body

application/json ·

Property	Type	Description
`session_id`	string	The session identifier returned by a previous call. Omit on the first call to create a new session and load a document.
`input`	string	The input document to load. Required when session_id is not provided. When input_format is 'Base64EncodedFile', provide a base64-encoded file payload. When 'FileIdentifier', provide the file ID previously returned by the /lmkit/v1/files/upload endpoint.
`input_format`		How the input is provided. Accepted values: 'Base64EncodedFile' or 'FileIdentifier'. Defaults to 'Base64EncodedFile'.
`question`	string	The user question to ask about the loaded document. Required when session_id is provided. Optional on the first call (if provided, the question is submitted immediately after loading the document).
`model`	string	The identifier of the chat model to use for response generation. If not provided, the server's default chat model is used. Only used when creating a new session.
`embedding_model`	string	The identifier of the embedding model to use for passage retrieval. If not provided, the server's default text embeddings model is used. Only used when creating a new session.
`stream`	boolean	When true, the response is streamed as server-sent events (SSE). Each event contains a JSON object with a 'delta' field for incremental text. Defaults to false.
`max_completion_tokens`	object (int32)	The maximum number of tokens permitted for the response. Set to -1 to disable the limit. Defaults to -1 (unlimited).

Responses

Status	Type	Description
200		OK
202		Accepted
400		Bad Request
404		Not Found
500	application/json	Internal Server Error

curl -X POST "$LMKIT_ONE_URL/lmkit/v1/chat-with-document" \
  -H "Authorization: Bearer $LMKIT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "session_id": "string",
  "input": "",
  "input_format": "Base64EncodedFile",
  "question": ""
}'

Skip the model's thinking phase and start generating the answer.#

POST/lmkit/v1/chat-with-document/{sessionId}/skip-thinking

Signals the model to immediately end its internal reasoning (chain-of-thought) and begin producing the user-visible answer. Has no effect if the model is not currently thinking.

Parameters

Name	In	Type	Description
`sessionId`required	path	string	The session identifier.

Responses

Status	Type	Description
200		OK
404		Not Found

curl -X POST "$LMKIT_ONE_URL/lmkit/v1/chat-with-document/$SESSIONID/skip-thinking" \
  -H "Authorization: Bearer $LMKIT_API_KEY"

Clear the conversation history while keeping the loaded document.#

POST/lmkit/v1/chat-with-document/{sessionId}/clear-history

Resets the chat history so you can start a fresh conversation without reloading the document.

Parameters

Name	In	Type	Description
`sessionId`required	path	string	The session identifier.

Responses

Status	Type	Description
200		OK
404		Not Found

curl -X POST "$LMKIT_ONE_URL/lmkit/v1/chat-with-document/$SESSIONID/clear-history" \
  -H "Authorization: Bearer $LMKIT_API_KEY"

Delete a chat session and release its resources.#

DELETE/lmkit/v1/chat-with-document/{sessionId}

Removes the chat session, releases the loaded models, and frees memory.

Parameters

Name	In	Type	Description
`sessionId`required	path	string	The session identifier to delete.

Responses

Status	Type	Description
200		OK
404		Not Found

curl -X DELETE "$LMKIT_ONE_URL/lmkit/v1/chat-with-document/$SESSIONID" \
  -H "Authorization: Bearer $LMKIT_API_KEY"