Documentation Index
Fetch the complete documentation index at: https://docs.api.ecylt.com/llms.txt
Use this file to discover all available pages before exploring further.
Wbot-4:347B is now live. Compared to the original Wbot-4, this version offers free API access.
API access: https://api.223387.xyz/
How to try Wbot-4-Preview-low?Answer: Visit https://api.223387.xyz/ to try it out!
Conversational application API
The conversational application supports session persistence, allowing previous chat history to serve as context for responses. It is suitable for chat/customer service AI and similar use cases.Base URL
Authentication
The Service API usesAPI-Key for authentication. We strongly recommend that developers store the API-Key on the backend rather than sharing it or storing it on the client side, to prevent API-Key leakage and potential financial loss. All API requests should include your API-Key in the Authorization HTTP Header, as shown below:
POST
Send a conversation message
Create a conversation message.Request Body
-
Name
queryTypestringDescription User input/question content. -
Name
inputsTypeobjectDescription (Optional) Allows passing values for App-defined variables. The inputs parameter contains multiple key/value pairs, where each key corresponds to a specific variable and each value is the concrete value for that variable. -
Name
response_modeTypestringDescriptionstreamingStreaming mode (recommended). Implements a typewriter-style streaming return based on SSE (Server-Sent Events).
-
Name
userTypestringDescription User identifier, used to define the identity of the end user for easy retrieval and statistics. Defined by the developer, must be unique within the application. -
Name
conversation_idTypestringDescription (Optional) Conversation ID. To continue a conversation based on previous chat history, the conversation_id from the previous message must be provided. -
Name
auto_generate_nameTypeboolDescription (Optional) Auto-generate title, defaults tofalse. You can asynchronously generate a title by calling the conversation rename API withauto_generateset totrue.
Response
- When
response_modeisstreaming, returns a ChunkChatCompletionResponse object stream sequence. -
ChunkChatCompletionResponse
- Returns streaming chunks of App output, with
Content-Typeastext/event-stream. Each streaming chunk starts with data:, and chunks are separated by \n\n (two newline characters), as shown below: -
event: messageLLM text block return event: the complete text is output in chunks.task_id(string) Task ID, used for request tracking and the stop response API belowmessage_id(string) Unique message IDconversation_id(string) Conversation IDanswer(string) LLM returned text block contentcreated_at(int) Creation timestamp, e.g.: 1705395332
event: agent_messageText block return event in Agent mode: in Agent mode, text is output in chunks (only used in Agent mode)task_id(string) Task ID, used for request tracking and the stop response API belowmessage_id(string) Unique message IDconversation_id(string) Conversation IDanswer(string) LLM returned text block contentcreated_at(int) Creation timestamp, e.g.: 1705395332
event: agent_thoughtContent related to Agent thinking steps in Agent mode, involving tool calls (only used in Agent mode)id(string) agent_thought ID, each Agent iteration round has a unique idtask_id(string) Task ID, used for request tracking and the stop response API belowmessage_id(string) Unique message IDposition(int) Position of agent_thought in the message, e.g., position is 1 for the first iterationthought(string) Agent’s thinking contentobservation(string) Return result of tool callstool(string) List of tools used, separated by ;tool_input(string) Tool input, JSON formatted string (object). E.g.:{"dalle3": {"prompt": "a cute cat"}}created_at(int) Creation timestamp, e.g.: 1705395332message_files(array[string]) File IDs associated with the currentagent_thoughtfile_id(string) File ID
conversation_id(string) Conversation ID
event: message_fileFile event, indicates a new file needs to be displayedid(string) Unique file IDtype(string) File type, currently only imagebelongs_to(string) File belongs to user or assistant, this API only returnsassistanturl(string) File access URLconversation_id(string) Conversation ID
event: message_endMessage end event, receiving this event means the streaming return has ended.task_id(string) Task ID, used for request tracking and the stop response API belowmessage_id(string) Unique message IDconversation_id(string) Conversation IDmetadata(object) Metadatausage(Usage) Model usage informationretriever_resources(array[RetrieverResource]) Citation and attribution segment list
event: message_replaceMessage content replacement event. When content moderation is enabled and the output triggers moderation rules, this event replaces the message content with the preset reply.task_id(string) Task ID, used for request tracking and the stop response API belowmessage_id(string) Unique message IDconversation_id(string) Conversation IDanswer(string) Replacement content (directly replaces all LLM reply text)created_at(int) Creation timestamp, e.g.: 1705395332
event: errorExceptions during the streaming output are output as stream events. The stream ends after receiving an error event.task_id(string) Task ID, used for request tracking and the stop response API belowmessage_id(string) Unique message IDstatus(int) HTTP status codecode(string) Error codemessage(string) Error message
event: pingPing event every 10 seconds to keep the connection alive.
Errors
- 404, conversation does not exist
- 400,
invalid_param, invalid parameter - 400,
app_unavailable, App configuration is unavailable - 400,
provider_not_initialize, no available model credential configuration - 400,
provider_quota_exceeded, model call quota exceeded - 400,
model_currently_not_support, current model is unavailable - 400,
completion_request_error, text generation failed - 500, internal server error