mcp-classify mcp

v1.0.0 · MCP Tool · ai · registry.pascalai.org

Classify text into predefined or custom categories using LLM-based zero-shot or few-shot classification via MakerAI. Returns ranked labels with confidence scores. Supports single-label and multi-label classification.

classifyclassificationlabelsnlpzero-shotcategorize

MakerAI Pipeline

pipeline: IPAILLMProvider.CompleteStructured — classification with JSON schema output

Input Parameters

Parameter	Type	Description
textrequired	string	Text to classify.
labelsrequired	array[string]	List of possible category labels.
multi_labeloptional	boolean	Allow multiple labels per text. Default: false. Default: `False`.
examplesoptional	array[object]	Few-shot examples to improve accuracy.
modeloptional	string	LLM model. Empty = router default.
thresholdoptional	number	Minimum confidence score to include a label (multi_label). Default: 0.5. Default: `0.5`.

Output Fields

Field	Type	Description
label	string	Top predicted label (single-label).
confidence	number
labels	array[object]	All labels with scores, ranked by confidence.
model	string

Examples

Classify customer support ticket

// Input
{
  "text": "My payment was charged twice for the same order!",
  "labels": [
    "billing",
    "shipping",
    "technical",
    "account",
    "other"
  ]
}

// Output
{
  "label": "billing",
  "confidence": 0.97,
  "labels": [
    {
      "label": "billing",
      "score": 0.97
    },
    {
      "label": "account",
      "score": 0.02
    }
  ],
  "model": "claude-sonnet-4-6"
}

Install & Discovery

Install

ppm install mcp-classify

Get JSON Schema

GET /v1/packages/mcp-classify/1.0.0/schema

Discover by keyword

GET /v1/mcp/discover?q=classify

Discovery hint: Install with ppm install mcp-classify or invoke remotely via POST /v1/invoke/mcp-classify on the MCP Service.

PascalAI Usage

uses toolslib;
var Tool := LoadTool('mcp-classify');
var R := Tool.Call(JsonObj(['text',ticketText,'labels',JsonArr(['billing','shipping','technical','other'])]));
Writeln(R['label'] + ' (' + FloatToStr(R['confidence']) + ')');