ai.classify

Operation: ai.classify
Category: ai
Tags: text, classification

Classify text into one of a set of labels using an AI provider

Runtime

Type: AI Provider (routed via provider registry)
Timeout: 300s
Retries: 3 (ExponentialWithJitter)

Inputs

Name	Type	Required	Default	Description
text	String	Yes	—	Text to classify

Outputs

Name	Type	Description
label	String	Selected label
confidence	String	Confidence level (high/low)
labels	Array<String>	All available labels
raw_response	String	Raw provider response

Default Configuration

{
  "labels": [
    "positive",
    "negative",
    "neutral"
  ]
}

Usage

TypeScript

import { WorkflowBuilder } from "@fabric-platform/sdk";

const workflow = new WorkflowBuilder("my-workflow")
  .node("ai-classify", "ai_invoke", (n) =>
    n.config({
      operation: "ai.classify",
      // ... node-specific config
    })
  )
  .build();

Python

from fabric_platform import FabricClient

fabric = FabricClient(api_key="fab_xxx")

wf_id = fabric.upsert_workflow("my-workflow", nodes=[
    {
        "key": "ai-classify",
        "kind": "ai_invoke",
    },
])

Rust

use fabric_sdk::FabricClient;

let client = FabricClient::new("http://localhost:3001", api_key)?;

let wf_id = client.upsert_workflow("my-workflow", serde_json::json!({
    "nodes": [{
        "key": "ai-classify",
        "kind": "ai_invoke"
    }]
})).await?;

curl

curl -X POST http://localhost:3001/v1/workflow-definitions \
  -H "Authorization: Bearer $FABRIC_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "my-workflow",
    "nodes": [{
      "key": "ai-classify",
      "kind": "ai_invoke"
    }]
  }'

Example

{
  "key": "classify-sentiment",
  "operation": "ai.classify",
  "config": {
    "labels": ["positive", "negative", "neutral"]
  },
  "inputs": [{ "name": "text", "path": "$context.review_text" }],
  "outputs": [{ "name": "label", "path": "$context.sentiment" }]
}

Common Errors

Classification returned unknown label

Cause: The AI response did not match any of the configured labels

Fix: The raw response is returned as the label with confidence ‘low’. Add more descriptive labels or use a prompt_template for better accuracy