mcp-summarize mcp

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

Summarize long documents, articles, conversations or code using LLM via MakerAI. Controls length, format (bullets, paragraphs, TL;DR), language, focus topics and handles documents exceeding context windows via chunked summarization.

summarizesummarytldrdocumentarticlellmabstract

MakerAI Pipeline

Input Parameters

ParameterTypeDescription
textoptional string Text or document to summarize.
urloptional string URL to fetch and summarize (alternative to text — uses mcp-fetch internally).
formatoptional string 'paragraph': prose summary; 'bullets': bullet point list; 'tldr': one-sentence summary; 'structured': title + key points + conclusion. Default: 'paragraph'. One of: paragraph, bullets, tldr, structured. Default: paragraph.
lengthoptional string Desired summary length. Default: 'medium'. One of: short, medium, long. Default: medium.
focusoptional array[string] Topics or aspects to focus on in the summary. E.g. ['pricing', 'technical requirements'].
languageoptional string Output language for the summary. Default: same as input. Default: .
modeloptional string
audienceoptional string Target audience hint (e.g. 'executive', 'technical', 'general public').

Output Fields

FieldTypeDescription
summary string Generated summary text.
title string Inferred document title (structured format).
key_points array[string] Main points (structured/bullets format).
word_count integer Word count of the summary.
source_tokens integer Token count of the original text.
chunks_processed integer Number of chunks for long documents.
model string
tokens_used integer

Examples

Summarize a research paper as bullet points

// Input
{
  "text": "Abstract: This paper presents...[10,000 words]",
  "format": "bullets",
  "length": "short",
  "focus": [
    "methodology",
    "results"
  ]
}

// Output
{
  "summary": "\u2022 Novel transformer architecture reduces inference cost by 40%\n\u2022 Evaluated on 12 benchmarks, outperforms SOTA in 9\n\u2022 Training data: 500B tokens from curated web corpus",
  "key_points": [
    "40% cost reduction",
    "SOTA on 9/12 benchmarks"
  ],
  "word_count": 38,
  "source_tokens": 8200,
  "chunks_processed": 4,
  "tokens_used": 1240
}

TL;DR of a URL

// Input
{
  "url": "https://example.com/blog/ai-trends-2025",
  "format": "tldr"
}

// Output
{
  "summary": "AI in 2025 is defined by multimodal models, agentic systems and on-device inference becoming mainstream.",
  "word_count": 18
}

Install & Discovery

Install

ppm install mcp-summarize

Get JSON Schema

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

Discover by keyword

GET /v1/mcp/discover?q=summarize
Discovery hint: Install with ppm install mcp-summarize or invoke remotely via POST /v1/invoke/mcp-summarize on the MCP Service.

PascalAI Usage

uses toolslib;
var Tool := LoadTool('mcp-summarize');
var R := Tool.Call(JsonObj(['text',longDoc,'format','bullets','length','short']));
Writeln(R['summary']);