FastGPT icon

FastGPT

AI智能客服平台

Actions37

对话 → 请求插件

AI-generated

Summary

Calls a FastGPT plugin-type app by POSTing to /v1/chat/completions with stream disabled, passing plugin inputs as variables.

Inputs

  • chatId (required) — Chat session ID sent as chatId in the request body. Defaults to n8n-chat-id.
  • variables — Key/value list of plugin input parameters. The node converts it to an object and sends it as variables in the body.
  • requestOptions — Node-level HTTP settings: batching (items per batch, interval in ms), ignore SSL certificate issues, proxy URL, and timeout in ms (default 10000).

Output shape

The parsed JSON response body from FastGPT's /v1/chat/completions endpoint, returned as the node's output item(s). The operation's description says plugin results are read from pluginData.

The body is {stream:false, chatId, variables}. No messages field is sent. The description says the endpoint defaults to detail mode, but the source doesn't set detail for this operation, so pluginData presence depends on the FastGPT server. Errors are thrown as NodeApiError on any HTTP 4xx/5xx status or when the body's code is not 200 (the message comes from the body's message). The node's HTTP defaults ignore status errors so this check handles them. The base URL comes from the fastGPTApi credential.

Examples

Example 1: Run a FastGPT plugin that takes a city name and returns weather data

Set chatId to a unique value, add variable key city with value Beijing, run the node, then read pluginData from the output.

Example 2: Throttle plugin calls when processing many input items

Under Request Options, enable Batching with Items per Batch 10 and Batch Interval 2000 ms, and raise Timeout for slow plugins.

Links

Discussion