Pustaka is an MCP server that searches thesis repositories across 75+ Indonesian universities — PTN, PTS, and Polytechnics — via OAI-PMH harvesting and HTML scraping. It detects which theses are truly full-text (Bab 1 through Daftar Pustaka) by analyzing PDF content directly.

This is a port of pencari-skripsi-indonesia (MIT) by Yingtze — original credit below.

Quick Start

# Install
pip install "pustaka-mcp[pdf]"

# Run (stdio transport)
pustaka

With uvx

uvx pustaka-mcp

Claude Desktop config

{
  "mcpServers": {
    "pustaka": {
      "command": "uvx",
      "args": ["pustaka-mcp"]
    }
  }
}

Tools

Resources

• skripsi://universities — all universities as JSON
• skripsi://stats — repository statistics

Status Meanings

Environment Variables

Requirements

• Python 3.10+
• pdfminer.six (optional, for PDF content analysis — recommended)

pip install pustaka-mcp[pdf]

Without pdfminer.six, the server works but uses HTML-only detection (less accurate).

Structure

├── src/pustaka/
│   ├── __init__.py         ← version
│   ├── models.py           ← data models + scoring
│   ├── pdf_analyzer.py     ← PDF content analysis
│   ├── engine.py           ← OAI-PMH + HTML scraping engine
│   └── server.py           ← FastMCP server
├── tests/
│   └── test_mcp.py         ← 35+ unit tests
├── repositories.json       ← 75+ university configs
├── .vscode/
│   └── PORT_FROM_PREVIOUS_REPO.md  ← port notes
├── pyproject.toml
├── LICENSE
└── README.md

Original Work

This project is a port of pencari-skripsi-indonesia by Yingtze, released under the MIT License. The original work provides the core repository configurations, OAI-PMH harvesting logic, HTML parsers, and PDF analysis heuristics that make this possible.

Without Yingtze's groundwork of cataloging 75+ Indonesian university repositories and building the scraping infrastructure, this MCP server would not exist.

License

AGPL-3.0-only — This port. See LICENSE.

The original pencari-skripsi-indonesia is MIT licensed.