journal-search-mcp

journal-search-mcp

MCP server for searching and verifying academic journal PDFs. It uses Tavily search and Gemini Flash Lite to analyze PDF content and return structured JSON results.

Category
Visit Server

README

journal-search-mcp

MCP Server (Node.js + TypeScript, Streamable HTTP transport) untuk pencarian dan verifikasi jurnal akademik PDF. Dipanggil oleh Claude sebagai MCP client/orchestrator. Claude tidak pernah membaca isi dokumen mentah — semua pembacaan/pemahaman konten PDF didelegasikan ke sub-agent Gemini Flash Lite yang mengembalikan JSON pendek.

Lihat specs.md (dokumen asli) untuk spesifikasi lengkap. README ini hanya ringkasan cara menjalankan.

Instalasi

npm install
cp .env.example .env
# isi .env: TAVILY_API_KEY, GEMINI_API_KEYS, MCP_API_KEY minimal

Menjalankan lokal

npm run build
npm start
# atau untuk development dengan watch mode:
npm run dev   # di terminal lain: node dist/index.js setiap kali rebuild

Server akan listen di http://localhost:3000 (atau MCP_PORT/PORT kalau di-set).

Cek cepat:

curl http://localhost:3000/health
curl "http://localhost:3000/test?key=ISI_MCP_API_KEY_ANDA"

Endpoint MCP untuk dipasang di Claude:

http://localhost:3000/mcp?key=ISI_MCP_API_KEY_ANDA

REST API (HTTP biasa)

Selain MCP, setiap tool juga bisa dipanggil lewat REST endpoint HTTP biasa — berguna untuk integrasi dengan backend non-MCP, Postman, curl, atau agent yang ingin akses JSON langsung tanpa protokol MCP.

Autentikasi

Sama seperti MCP endpoint: API key dikirim lewat query string ?key=....

POST /api/verify-pdf?key=ISI_MCP_API_KEY_ANDA

Kalau key salah atau tidak ada → 401 { "error": "invalid_or_missing_api_key" }.

Daftar endpoint

Method Path Deskripsi
POST /api/tavily-search Pencarian web umum (Tavily Search)
POST /api/tavily-extract Ekstraksi konten halaman web
POST /api/tavily-crawl Crawl/jelajahi situs
POST /api/verify-pdf Verifikasi magic bytes PDF dari URL
POST /api/analyze-journal-standard Analisis jurnal standar (worker round-robin)
POST /api/analyze-journal-custom Analisis jurnal dengan prompt/field kustom
POST /api/search-and-check-journals Tool komposit: search → verify → judge
GET /api/tools Daftar semua tool yang tersedia (metadata)

Semua endpoint POST menerima Content-Type: application/json. Input body mengikuti schema yang sama dengan tool MCP — lihat src/schemas/toolSchemas.ts.

Contoh request

# Verifikasi PDF
curl -X POST "http://localhost:3000/api/verify-pdf?key=ISI_MCP_API_KEY_ANDA" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com/paper.pdf"}'

Response sukses (200):

{
  "url": "https://example.com/paper.pdf",
  "is_pdf": true,
  "detected_signature": "%PDF-",
  "http_status": 200
}

Response validasi gagal (400):

{
  "error": "invalid_input",
  "error_detail": "url: url harus valid"
}

Response tanpa/ salah API key (401):

{
  "error": "invalid_or_missing_api_key"
}

Discovery

curl "http://localhost:3000/api/tools?key=ISI_MCP_API_KEY_ANDA"

Mengembalikan daftar 7 tool dengan nama, method HTTP, dan path masing-masing.

Tools yang tersedia

Tool Deskripsi
tavily_search Pass-through pencarian web umum
tavily_extract Pass-through ekstraksi konten halaman web
tavily_crawl Pass-through crawl situs
verify_pdf Cek magic bytes URL — benar PDF atau bukan
analyze_journal_standard verify → extract → judge (worker gemini-1/2 round-robin)
analyze_journal_custom Sama seperti standard, tapi worker gemini-3 + custom prompt opsional
search_and_check_journals Tool komposit utama: search → verify → judge, banyak kandidat sekaligus

Error handling

Semua tool tidak pernah throw ke MCP client — setiap kegagalan (download gagal, bukan PDF, semua API key rate limited, JSON tidak valid dari Gemini, dll) dibungkus jadi field error / error_detail pada JSON output. Lihat kode error lengkap di src/types.ts (ToolErrorCode) dan §11 specs.md.

Level HTTP (di luar tool JSON):

  • 401 invalid_or_missing_api_key
  • 500 server_misconfigured_missing_mcp_api_key
  • 400 invalid_json_body
  • 404 not_found

Testing

npm run typecheck

Lihat §12 specs.md untuk testing plan lengkap (unit test rate limiter, key pool, magic bytes checker, integration test dengan PDF publik, dll — belum di-generate otomatis di sini, tulis sesuai kebutuhan Anda dengan test runner pilihan, mis. vitest).

Deploy ke Render

Lihat render.yaml (infra-as-code) dan §14 specs.md. Ringkas:

  • Type: Web Service
  • Build: npm install && npm run build
  • Start: node dist/index.js
  • Health check path: /health
  • PORT otomatis di-inject Render — jangan hardcode port lain.
  • Set semua secret (TAVILY_API_KEY, GEMINI_API_KEYS, MCP_API_KEY) manual lewat Render Dashboard → Environment (jangan commit ke repo).

Non-goals fase ini

  • Worker 4 (Lightpanda/browser-rendering) — slot sudah disiapkan di workerPool.ts, tapi memanggilnya akan return error not_implemented.
  • Caching hasil analisis.
  • OCR untuk PDF hasil scan.
  • Multi-user auth (hanya satu shared MCP_API_KEY).

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured