Skip to main content
Glama

20 Best PDF MCP Servers, Compared (September 2026)

Updated 20 of 1,318 PDF servers in the Glama registryRebuilt weekly

The short answer

For most use cases, pdf-mcp (jztan/pdf-mcp) is the server to reach for: it exposes 13 tools graded A, had a commit 0 days ago, and 811 commits in the last 12 weeks, covering reading, search, and extraction from large PDFs. If you need a one-command npm install and are willing to accept a single read tool, PDF Reader MCP Server (SylphxAI/pdf-reader-mcp) is a simpler fallback, but it only exposes one tool and had zero npm downloads in a typical week. For privacy-first on-device search over many documents, Local RAG (shinpr/mcp-local-rag) offers 9 tools, an npm package, and 2,657 weekly downloads.

Whichever you choose, give it the narrowest access that still works (a read-only credential, a replica, a scratch account), and widen it only once you have watched what your agent actually asks for.

Glama operates the MCP registry these numbers are measured from, and sells MCP hosting and a gateway. No position on this page is paid for. How the registry is built.

Quick picks

  • 1PDF Reader MCP Server : For local PDF text and metadata extraction in an agent: it exposes read_pdf for local or URL sources with page selection, and installs via npx.
  • 2kordoc : South Korean public-institution document parsing: it turns HWP, HWPX and PDF into Markdown with table reconstruction and hardened extraction.
  • 3Local RAG : Privacy-first local document search: it runs entirely on-device with no cloud services and exposes ingest, hybrid semantic/keyword query, and sync tools.
  • 4pdf-mcp : Situations that demand reading only the relevant pages from large PDFs or folders: hybrid search plus corpus triage avoids loading everything.
  • 5Docling MCP : Teams running Docling Serve who need PDF to structured JSON: remote conversion mode exposes the conversion tools and supports local fallback.

Which one, for your situation

Your situationWhat to use
Need to read and search large PDFs with many toolsUse pdf-mcp (jztan/pdf-mcp) for 13 tools, A grade, and a commit 0 days ago.
Need to fill and sign PDFs locallyUse PDF Tools (Open-Document-Alliance/PDF-Tools) for local form filling and signing without uploading files.
Need to convert documents between formatsUse mcp-pandoc (vivekVells/mcp-pandoc) to convert between Markdown, HTML, DOCX, and PDF with one tool.
Need OCR for scanned or complex documentsUse PaddleOCR (PaddlePaddle/PaddleOCR) for OCR, handwriting recognition, and structure-aware parsing.
Need privacy-first local search across documentsUse Local RAG (shinpr/mcp-local-rag) for on-device semantic search with 9 tools.

Top MCP servers for PDF

Best forProfile
1
SylphxAI/pdf-reader-mcp
For local PDF text and metadata extraction in an agent: it exposes read_pdf for local or URL sources with page selection, and installs via npx.Community favourite905+25yesterday78.7
2
chrisryugj/kordoc
South Korean public-institution document parsing: it turns HWP, HWPX and PDF into Markdown with table reconstruction and hardened extraction.Community favourite1,781+143yesterday73.6
3
shinpr/mcp-local-rag
Privacy-first local document search: it runs entirely on-device with no cloud services and exposes ingest, hybrid semantic/keyword query, and sync tools.Community favourite371+17yesterday68.5
4
jztan/pdf-mcp
Situations that demand reading only the relevant pages from large PDFs or folders: hybrid search plus corpus triage avoids loading everything.Emerging123+35today64.0
5
docling-project/docling-mcp
Teams running Docling Serve who need PDF to structured JSON: remote conversion mode exposes the conversion tools and supports local fallback.Community favourite725+297 days ago62.3
6
vivekVells/mcp-pandoc
Converting documents between Markdown, HTML, DOCX, and PDF when one tool must cover many formats: it exposes a single convert-contents tool with Pandoc behind it.Community favourite579+915 days ago59.9
7
zcaceres/markdownify-mcp
A one-afternoon install for converting PDF, Office, image, audio, and web content to Markdown: it exposes ten conversion tools, including pdf-to-markdown.Abandoned but popular2,911+106121 days ago59.9
8
Open-Document-Alliance/PDF-Tools
Local PDF form filling and signing in Claude Desktop: the README documents form filling, signature zones, and saved local signatures without uploading files.Steady152+47 days ago59.4
9
claude-world/notebooklm-skill
Turning PDFs and URLs into NotebookLM notebooks from a terminal or MCP client: 13 tools cover creation, grounded Q&A, summaries, and artifact downloads.Community favourite443+4343 days ago58.8
10
VectifyAI/pageindex-mcp
Long PDFs on MCP-compatible clients without vector database infrastructure: it exposes a reasoning-based tree-structured index that lets agents navigate content like a human.Community favourite382+636 days ago58.2
11
I-CAN-hack/pdf-mcp
LLMs needing to pull text, tables, and diagrams from datasheets and technical PDFs: it renders pages to images and extracts text in multiple formats.Emerging66+3032 days ago58.0
12
chrisryugj/korean-dart-mcp
Someone analyzing Korean DART filings with an AI assistant: this server exposes 15 tools for disclosures, financials, XBRL, and HWP/PDF attachment conversion to markdown.Emerging92+931 days ago57.4
13
Agents365-ai/zotero-cli-ai
Zotero users who want an agent to search and extract PDF text from a local library: it reads local SQLite directly for offline, millisecond responses.Steady202+9today57.3
14
PaddlePaddle/PaddleOCR
RAG pipelines that need complex PDFs and images converted to Markdown or JSON: it offers OCR, handwriting recognition, and structure-aware parsing.Community favourite88,510+1,74539 days ago55.3
15
yusufkaraaslan/Skill_Seekers
Turning docs, repos, PDFs, and videos into Claude skills and RAG knowledge: it exposes scraping tools plus packaging and vector database export.Community favourite14,864+18821 days ago54.3
16
lfnovo/content-core
Extracting text from PDFs, web pages, and videos with no API key for most sources: extract_content auto-selects engines and returns structured JSON.Steady169+231 days ago53.0
17
opendatalab/MinerU-Document-Explorer
Building a literature knowledge base from a folder of PDFs: its Retrieve, Deep Read, and Ingest suites let an agent search, navigate, and ingest documents.Abandoned but popular616+17130 days ago51.0
18
Kentucky-ai/opentakeoff
For agent-driven construction takeoffs from plan PDFs: 42 MCP tools cover scaling, room measurement, and DXF/PDF exports.Emerging101+46yesterday50.4
19
KDAN-PDF/KDAN-PDF-MCP
For chat-based PDF cleanup before sharing: a remote MCP connector for compressing, removing pages, redacting, protecting, and comparing PDFs in Claude or ChatGPT.Steady57013 days ago49.6
20
enconvert/mcp
Teams needing file conversion plus live-web data work: it exposes conversion, rendering, search, extraction, ingest, and page-monitoring tools.Emerging7No snapshot history24 days ago48.6

The ranking, with the evidence

Each position is a weighted mean of adoption (40%), maintenance (24%), momentum (14%), tool description quality (13%) and trust (9%), multiplied by three attenuators: how directly the server is about PDF (named for it, declaring it, tagged with it, or merely mentioning it), whether its repository is still moving, and how much independent evidence of adoption it has. Open the score on any entry to see every number, including the ones marked ≈, which were imputed from the median of the other candidates rather than measured. The maintenance grade on each entry is mostly issue responsiveness, release recency and open security alerts rather than commits, so a recent commit beside a low grade is two different measurements rather than a contradiction.

  • Abandoned but popular: People use it, but its default branch has stopped moving. Fine to keep running, risky to adopt.
  • Community favourite: Widely adopted and still actively maintained.
  • Dormant: Neither changing nor widely adopted. Here because it still matches the search.
  • Emerging: Small audience, growing quickly, maintained. The bet with the most upside.
  • Steady: Maintained, modest audience, no surprises in either direction.
  • 1

    PDF Reader MCP Server

    Community favouriteNo change
    SylphxAI/pdf-reader-mcp ↗

    Best for: For local PDF text and metadata extraction in an agent: it exposes read_pdf for local or URL sources with page selection, and installs via npx.

    It exposes one MCP tool, read_pdf, which reads text, metadata, and page count from one or more PDFs at local paths or URLs and can limit extraction to selected pages. It depends on a platform-specific native package and fails closed if that binary is missing, so check that the host OS and architecture appear in the README's platform list before relying on it.

    GitHub stars
    905
    Stars over the last 90 days: 777 at the earliest of 38 readings and 915 at the latest, up 138.
    Stars / 30 days
    +25
    npm / typical week
    0
    Tools exposed
    1
    Last commit
    yesterday
    Commits / 12 weeks
    407
    Weekly commits over the last 12 weeks, oldest first: 23, 15, 23, 22, 21, 196, 37, 40, 18, 5, 1, 6. 407 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    B
    Score 78.7: show every number behind it
    • Adoption74 / 100 · weight 40%
      • GitHub stars74
      • npm downloads0
        downloads show none of the weekday rhythm human traffic has; halved
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance100 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence100
    • Momentum56 / 100 · weight 14%
      • Stars gained, relative to size63
      • Stars gained, absolute57
      • npm download trend44
    • Tool quality63 / 100 · weight 13%
      • Tool description quality55
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    78.7
    × relevance: the keyword is dedicated here
    1.00
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    78.7
  • 2

    kordoc

    Community favouriteNo change
    chrisryugj/kordoc ↗

    Best for: South Korean public-institution document parsing: it turns HWP, HWPX and PDF into Markdown with table reconstruction and hardened extraction.

    The README documents an MCP server and CLI that converts HWP, HWPX, PDF, Office and image files to Markdown, and also supports generating, patching, filling, redacting and OCR of documents. Before choosing it, note that it requires Node.js 18+ and is installed by running npx -y kordoc setup.

    GitHub stars
    1,781
    Stars over the last 90 days: 1,133 at the earliest of 33 readings and 1,792 at the latest, up 659.
    Stars / 30 days
    +143
    npm / typical week
    12K
    Tools exposed
    never inspected
    Last commit
    yesterday
    Commits / 12 weeks
    248
    Weekly commits over the last 12 weeks, oldest first: 4, 12, 76, 63, 22, 12, 11, 20, 3, 6, 4, 5. 238 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    Not graded
    Score 73.6: show every number behind it
    • Adoption99 / 100 · weight 40%
      • GitHub stars81
      • npm downloads87
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance100 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence100
    • Momentum88 / 100 · weight 14%
      • Stars gained, relative to size100
      • Stars gained, absolute87
      • npm download trend68
    • Tool quality≈73 / 100 · weight 13%
      • Tool description quality≈73
        tool descriptions not yet scored
      • Built and inspected by Glamanot measured
        never built and inspected by Glama
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    94.4
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    73.6
  • 3

    Local RAG

    Community favouriteNo change
    shinpr/mcp-local-rag ↗

    Best for: Privacy-first local document search: it runs entirely on-device with no cloud services and exposes ingest, hybrid semantic/keyword query, and sync tools.

    The server indexes PDF, DOCX, TXT, and Markdown files from configured base directories and exposes nine tools for ingest, hybrid semantic/keyword search, context reads, deletion, and sync. It requires Node.js 22 or later, and the first run needs internet access to download the package and the default embedding model.

    GitHub stars
    371
    Stars over the last 90 days: 309 at the earliest of 17 readings and 384 at the latest, up 75.
    Stars / 30 days
    +17
    npm / typical week
    2.7K
    Tools exposed
    9
    Last commit
    yesterday
    Commits / 12 weeks
    174
    Weekly commits over the last 12 weeks, oldest first: 15, 15, 2, 18, 2, 0, 41, 4, 5, 3, 15, 6. 126 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    A
    Score 68.5: show every number behind it
    • Adoption83 / 100 · weight 40%
      • GitHub stars64
      • npm downloads73
      • Used through Glama41
    • Maintenance100 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence100
    • Momentum74 / 100 · weight 14%
      • Stars gained, relative to size75
      • Stars gained, absolute51
      • npm download trend100
    • Tool quality88 / 100 · weight 13%
      • Tool description quality80
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    87.8
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    68.5
  • 4

    pdf-mcp

    EmergingNo change
    jztan/pdf-mcp ↗

    Best for: Situations that demand reading only the relevant pages from large PDFs or folders: hybrid search plus corpus triage avoids loading everything.

    pdf-mcp exposes 13 tools for inspecting PDFs, searching within them, reading pages, running corpus-wide queries, rendering pages, extracting charts, and managing a cache. Before choosing it, note that OCR on scanned PDFs requires installing Tesseract on the system.

    GitHub stars
    123
    Stars over the last 90 days: 55 at the earliest of 38 readings and 130 at the latest, up 75.
    Stars / 30 days
    +35
    npm / typical week
    Ships no npm package
    Tools exposed
    13
    Last commit
    today
    Commits / 12 weeks
    811
    Weekly commits over the last 12 weeks, oldest first: 84, 48, 54, 40, 82, 45, 120, 82, 73, 15, 108, 157. 908 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    A
    Score 64.0: show every number behind it
    • Adoption52 / 100 · weight 40%
      • GitHub stars52
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance100 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence100
    • Momentum85 / 100 · weight 14%
      • Stars gained, relative to size100
      • Stars gained, absolute63
      • npm download trendnot measured
        no npm download history
    • Tool quality73 / 100 · weight 13%
      • Tool description quality65
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    75.3
    × relevance: the keyword is dedicated here
    1.00
    × continuity: actively changing
    1.00
    × evidence: modest but real audience
    0.85
    Composite score
    64.0
  • 5

    Docling MCP

    Community favourite▲ 1
    docling-project/docling-mcp ↗

    Best for: Teams running Docling Serve who need PDF to structured JSON: remote conversion mode exposes the conversion tools and supports local fallback.

    The README describes a document processing service with tools for PDF to structured JSON conversion, document generation, and local caching, and it accepts local files and URLs as sources. Before choosing it, note that remote mode requires access to a Docling Serve API, while local mode requires the local installation and DOCLING_MCP_CONVERSION_MODE setting.

    GitHub stars
    725
    Stars over the last 90 days: 658 at the earliest of 13 readings and 727 at the latest, up 69.
    Stars / 30 days
    +29
    npm / typical week
    Ships no npm package
    Tools exposed
    never inspected
    Last commit
    7 days ago
    Commits / 12 weeks
    23
    Maintenance grade
    A
    Tool descriptions
    Not graded
    Score 62.3: show every number behind it
    • Adoption72 / 100 · weight 40%
      • GitHub stars72
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance97 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence85
    • Momentum69 / 100 · weight 14%
      • Stars gained, relative to size75
      • Stars gained, absolute59
      • npm download trendnot measured
        no npm download history
    • Tool quality≈73 / 100 · weight 13%
      • Tool description quality≈73
        tool descriptions not yet scored
      • Built and inspected by Glamanot measured
        never built and inspected by Glama
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integrates100
    Weighted mean of the five
    79.9
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    62.3
  • 6

    mcp-pandoc

    Community favourite▲ 3
    vivekVells/mcp-pandoc ↗

    Best for: Converting documents between Markdown, HTML, DOCX, and PDF when one tool must cover many formats: it exposes a single convert-contents tool with Pandoc behind it.

    The server exposes one tool, convert-contents, which transforms input content or files between supported formats such as Markdown, HTML, DOCX, and PDF. Before choosing it, note that PDF support is write-only and still under development, so PDF input is not available and an output file path is required for PDF output.

    GitHub stars
    579
    Stars over the last 90 days: 554 at the earliest of 84 readings and 579 at the latest, up 25.
    Stars / 30 days
    +9
    npm / typical week
    Ships no npm package
    Tools exposed
    1
    Last commit
    15 days ago
    Commits / 12 weeks
    23
    Weekly commits over the last 12 weeks, oldest first: 0, 0, 0, 0, 0, 0, 5, 0, 7, 11, 0, 0. 23 in total, and the last figure is the current week so far.
    Maintenance grade
    B
    Tool descriptions
    A
    Score 59.9: show every number behind it
    • Adoption69 / 100 · weight 40%
      • GitHub stars69
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance92 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade80
      • Commit cadence85
    • Momentum43 / 100 · weight 14%
      • Stars gained, relative to size44
      • Stars gained, absolute40
      • npm download trendnot measured
        no npm download history
    • Tool quality93 / 100 · weight 13%
      • Tool description quality85
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    76.8
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    59.9
  • 7

    Markdownify MCP Server

    Abandoned but popular▼ 2
    zcaceres/markdownify-mcp ↗

    Best for: A one-afternoon install for converting PDF, Office, image, audio, and web content to Markdown: it exposes ten conversion tools, including pdf-to-markdown.

    The server exposes ten tools that transform PDFs, DOCX, XLSX, PPTX, images, audio, YouTube videos, Bing searches, and web pages into Markdown, plus get-markdown-file for existing Markdown files. The thing worth knowing before choosing it: the published Docker image installs only markitdown[pdf], so audio-to-markdown and image-to-markdown require a local install with the [all] extras.

    GitHub stars
    2,911
    Stars over the last 90 days: 2,756 at the earliest of 17 readings and 2,983 at the latest, up 227.
    Stars / 30 days
    +106
    npm / typical week
    downloads not counted
    Tools exposed
    10
    Last commit
    121 days ago
    Commits / 12 weeks
    0
    Maintenance grade
    C
    Tool descriptions
    A
    Score 59.9: show every number behind it
    • Adoption87 / 100 · weight 40%
      • GitHub stars87
      • npm downloadsnot measured
        npm names no repository for mcp-markdownify-server, so its downloads cannot be attributed
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance52 / 100 · weight 24%
      • Last commit touching this server68
        dated from the last commit on the default branch, re-read from GitHub at publication; github.com shows a push 0 days ago, which counts every ref; the stored date would have published 20 days ago
      • Repository maintenance grade55
      • Commit cadence5
    • Momentum79 / 100 · weight 14%
      • Stars gained, relative to size78
      • Stars gained, absolute82
      • npm download trendnot measured
        no npm download history
    • Tool quality73 / 100 · weight 13%
      • Tool description quality65
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    76.8
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    59.9
  • 8

    PDF Tools

    Steady▼ 1
    Open-Document-Alliance/PDF-Tools ↗

    Best for: Local PDF form filling and signing in Claude Desktop: the README documents form filling, signature zones, and saved local signatures without uploading files.

    The README documents a local PDF toolset for Claude Desktop and other MCP hosts: fill, sign, merge, split, extract, render, and analyze PDFs, plus an interactive viewer, URL-to-PDF download, reusable form profiles, and local signature zones. The one thing to know before choosing it: it targets local MCP hosts only and does not yet include a remote connector for web-hosted Claude Cowork.

    GitHub stars
    152
    Stars over the last 90 days: 140 at the earliest of 25 readings and 153 at the latest, up 13.
    Stars / 30 days
    +4
    npm / typical week
    downloads not counted
    Tools exposed
    never inspected
    Last commit
    7 days ago
    Commits / 12 weeks
    851
    Weekly commits over the last 12 weeks, oldest first: 0, 0, 0, 0, 0, 314, 132, 265, 117, 49, 19, 68. 964 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    Not graded
    Score 59.4: show every number behind it
    • Adoption55 / 100 · weight 40%
      • GitHub stars55
      • npm downloadsnot measured
        npm names no repository for pdf-tools, so its downloads cannot be attributed
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance100 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence100
    • Momentum40 / 100 · weight 14%
      • Stars gained, relative to size48
      • Stars gained, absolute29
      • npm download trendnot measured
        no npm download history
    • Tool quality≈73 / 100 · weight 13%
      • Tool description quality≈73
        tool descriptions not yet scored
      • Built and inspected by Glamanot measured
        never built and inspected by Glama
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    69.9
    × relevance: the keyword is dedicated here
    1.00
    × continuity: actively changing
    1.00
    × evidence: modest but real audience
    0.85
    Composite score
    59.4
  • 9

    notebooklm-skill

    Community favourite▼ 1
    claude-world/notebooklm-skill ↗

    Best for: Turning PDFs and URLs into NotebookLM notebooks from a terminal or MCP client: 13 tools cover creation, grounded Q&A, summaries, and artifact downloads.

    The server exposes 13 tools for NotebookLM notebook CRUD, mixed-source ingestion, grounded questions, summaries, artifact generation/list/download, and research pipelines, all through a JSON-first CLI and FastMCP server. It is an unofficial integration with NotebookLM's web API, so Google can change availability, quotas, or artifact behavior without notice.

    GitHub stars
    443
    Stars over the last 90 days: 252 at the earliest of 10 readings and 443 at the latest, up 191.
    Stars / 30 days
    +43
    npm / typical week
    Ships no npm package
    Tools exposed
    13
    Last commit
    43 days ago
    Commits / 12 weeks
    1
    Weekly commits over the last 12 weeks, oldest first: 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0. 1 in total, and the last figure is the current week so far.
    Maintenance grade
    B
    Tool descriptions
    B
    Score 58.8: show every number behind it
    • Adoption66 / 100 · weight 40%
      • GitHub stars66
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance83 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade80
      • Commit cadence40
    • Momentum86 / 100 · weight 14%
      • Stars gained, relative to size100
      • Stars gained, absolute66
      • npm download trendnot measured
        no npm download history
    • Tool quality61 / 100 · weight 13%
      • Tool description quality53
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    75.4
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    58.8
  • 10

    PageIndex MCP

    Community favourite▲ 1
    VectifyAI/pageindex-mcp ↗

    Best for: Long PDFs on MCP-compatible clients without vector database infrastructure: it exposes a reasoning-based tree-structured index that lets agents navigate content like a human.

    PageIndex MCP runs as an HTTP MCP server with API key authentication, and a local npx server is available for uploading local PDFs; both connect MCP clients to PageIndex's vectorless, tree-structured document index. The requirement to know: the hosted endpoint needs a PageIndex API key, and local PDF upload needs Node.js 18 or newer.

    GitHub stars
    382
    Stars over the last 90 days: 360 at the earliest of 10 readings and 382 at the latest, up 22.
    Stars / 30 days
    +6
    npm / typical week
    105
    Tools exposed
    never inspected
    Last commit
    36 days ago
    Commits / 12 weeks
    6
    Weekly commits over the last 12 weeks, oldest first: 0, 0, 0, 0, 0, 0, 6, 0, 0, 0, 0, 0. 6 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    Not graded
    Score 58.2: show every number behind it
    • Adoption68 / 100 · weight 40%
      • GitHub stars65
      • npm downloads22
        downloads show none of the weekday rhythm human traffic has; halved
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance93 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence65
    • Momentum48 / 100 · weight 14%
      • Stars gained, relative to size43
      • Stars gained, absolute34
      • npm download trend74
    • Tool quality≈73 / 100 · weight 13%
      • Tool description quality≈73
        tool descriptions not yet scored
      • Built and inspected by Glamanot measured
        never built and inspected by Glama
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    74.6
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    58.2
  • 11

    pdf-mcp

    Emerging▼ 1
    I-CAN-hack/pdf-mcp ↗

    Best for: LLMs needing to pull text, tables, and diagrams from datasheets and technical PDFs: it renders pages to images and extracts text in multiple formats.

    It exposes five filename-based tools: metadata and table of contents lookup, text extraction in json, text, markdown, or html, page rendering as PNG, and text search, with all requests stateless. It assumes uvx is installed, because the documented setup runs the server via uvx from the GitHub repository.

    GitHub stars
    66
    Stars over the last 90 days: 30 at the earliest of 8 readings and 77 at the latest, up 47.
    Stars / 30 days
    +30
    npm / typical week
    Ships no npm package
    Tools exposed
    5
    Last commit
    32 days ago
    Commits / 12 weeks
    1
    Weekly commits over the last 12 weeks, oldest first: 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0. 1 in total, and the last figure is the current week so far.
    Maintenance grade
    C
    Tool descriptions
    A
    Score 58.0: show every number behind it
    • Adoption46 / 100 · weight 40%
      • GitHub stars46
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance77 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade55
      • Commit cadence40
    • Momentum84 / 100 · weight 14%
      • Stars gained, relative to size100
      • Stars gained, absolute60
      • npm download trendnot measured
        no npm download history
    • Tool quality83 / 100 · weight 13%
      • Tool description quality75
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    68.2
    × relevance: the keyword is dedicated here
    1.00
    × continuity: actively changing
    1.00
    × evidence: modest but real audience
    0.85
    Composite score
    58.0
  • 12

    Korean DART MCP

    EmergingNew
    chrisryugj/korean-dart-mcp ↗

    Best for: Someone analyzing Korean DART filings with an AI assistant: this server exposes 15 tools for disclosures, financials, XBRL, and HWP/PDF attachment conversion to markdown.

    It exposes 15 tools that query Korean DART disclosures, financials, XBRL, shareholding data, and insider signals, and it converts HWP/PDF attachments to markdown. Choosing it assumes the Korean OpenDART disclosure system and requires Node.js 20.19+, so it is not a general-purpose PDF server.

    GitHub stars
    92
    Stars over the last 90 days: 51 at the earliest of 9 readings and 96 at the latest, up 45.
    Stars / 30 days
    +9
    npm / typical week
    203
    Tools exposed
    15
    Last commit
    31 days ago
    Commits / 12 weeks
    7
    Weekly commits over the last 12 weeks, oldest first: 1, 0, 0, 1, 0, 0, 4, 1, 0, 0, 0, 0. 7 in total, and the last figure is the current week so far.
    Maintenance grade
    B
    Tool descriptions
    A
    Score 57.4: show every number behind it
    • Adoption57 / 100 · weight 40%
      • GitHub stars49
      • npm downloads49
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance88 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade80
      • Commit cadence65
    • Momentum72 / 100 · weight 14%
      • Stars gained, relative to size82
      • Stars gained, absolute41
      • npm download trend90
    • Tool quality83 / 100 · weight 13%
      • Tool description quality75
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    73.6
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    57.4
  • 13

    zotero-cli-cc

    Steady▼ 1
    Agents365-ai/zotero-cli-ai ↗

    Best for: Zotero users who want an agent to search and extract PDF text from a local library: it reads local SQLite directly for offline, millisecond responses.

    It runs as an MCP server for Zotero reference management, reading via local SQLite, writing through the Zotero Web API, and extracting PDF text with ranked full-text search. The one requirement to know before choosing it is that write operations need a Zotero Web API key.

    GitHub stars
    202
    Stars over the last 90 days: 174 at the earliest of 11 readings and 202 at the latest, up 28.
    Stars / 30 days
    +9
    npm / typical week
    Ships no npm package
    Tools exposed
    never inspected
    Last commit
    today
    Commits / 12 weeks
    34
    Weekly commits over the last 12 weeks, oldest first: 2, 0, 0, 0, 10, 3, 0, 2, 0, 12, 5, 6. 40 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    Not graded
    Score 57.3: show every number behind it
    • Adoption58 / 100 · weight 40%
      • GitHub stars58
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance100 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence100
    • Momentum56 / 100 · weight 14%
      • Stars gained, relative to size67
      • Stars gained, absolute41
      • npm download trendnot measured
        no npm download history
    • Tool quality≈73 / 100 · weight 13%
      • Tool description quality≈73
        tool descriptions not yet scored
      • Built and inspected by Glamanot measured
        never built and inspected by Glama
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    73.5
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    57.3
  • 14

    PaddleOCR MCP Server

    Community favourite▼ 1
    PaddlePaddle/PaddleOCR ↗

    Best for: RAG pipelines that need complex PDFs and images converted to Markdown or JSON: it offers OCR, handwriting recognition, and structure-aware parsing.

    The server's README describes PDF and image parsing into structured Markdown and JSON, with OCR for text, handwriting, and complex documents. No npm package is published, and its exposed tool list is not measured, so verify how it is installed and what tools it offers before choosing it.

    GitHub stars
    88,510
    Stars over the last 90 days: 81,625 at the earliest of 14 readings and 88,510 at the latest, up 6,885.
    Stars / 30 days
    +1,745
    npm / typical week
    Ships no npm package
    Tools exposed
    never inspected
    Last commit
    39 days ago
    Commits / 12 weeks
    19
    Weekly commits over the last 12 weeks, oldest first: 13, 1, 3, 0, 0, 0, 2, 0, 0, 0, 0, 0. 19 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    Not graded
    Score 55.3: show every number behind it
    • Adoption100 / 100 · weight 40%
      • GitHub stars100
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance97 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence85
    • Momentum75 / 100 · weight 14%
      • Stars gained, relative to size58
      • Stars gained, absolute100
      • npm download trendnot measured
        no npm download history
    • Tool quality≈73 / 100 · weight 13%
      • Tool description quality≈73
        tool descriptions not yet scored
      • Built and inspected by Glamanot measured
        never built and inspected by Glama
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integrates100
    Weighted mean of the five
    92.2
    × relevance: the keyword is tagged here
    0.60
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    55.3
  • 15

    Skill Seekers

    Community favourite▼ 1
    yusufkaraaslan/Skill_Seekers ↗

    Best for: Turning docs, repos, PDFs, and videos into Claude skills and RAG knowledge: it exposes scraping tools plus packaging and vector database export.

    It exposes 40 MCP tools that cover config generation, scraping of docs, GitHub, PDFs, video, and codebases, plus packaging, upload, and export to vector databases. The one-command workflow makes AI enhancement mandatory, and API-based enhancement requires a platform API key, though a local mode can run without one.

    GitHub stars
    14,864
    Stars over the last 90 days: 14,058 at the earliest of 11 readings and 14,864 at the latest, up 806.
    Stars / 30 days
    +188
    npm / typical week
    Ships no npm package
    Tools exposed
    40
    Last commit
    21 days ago
    Commits / 12 weeks
    50
    Weekly commits over the last 12 weeks, oldest first: 3, 3, 0, 3, 0, 15, 1, 17, 8, 0, 0, 0. 50 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    B
    Score 54.3: show every number behind it
    • Adoption100 / 100 · weight 40%
      • GitHub stars100
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance100 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence100
    • Momentum64 / 100 · weight 14%
      • Stars gained, relative to size46
      • Stars gained, absolute92
      • npm download trendnot measured
        no npm download history
    • Tool quality66 / 100 · weight 13%
      • Tool description quality57
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    90.5
    × relevance: the keyword is tagged here
    0.60
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    54.3
  • 16

    content-core

    Steady▼ 1
    lfnovo/content-core ↗

    Best for: Extracting text from PDFs, web pages, and videos with no API key for most sources: extract_content auto-selects engines and returns structured JSON.

    The MCP server exposes extract_content, which pulls text from URLs, PDFs, Word docs, YouTube transcripts, audio, and video, and summarize_content, which sends text to an LLM. Extraction works without a key for most sources, but summarization requires an LLM provider key such as OPENAI_API_KEY to be configured.

    GitHub stars
    169
    Stars over the last 90 days: 158 at the earliest of 10 readings and 169 at the latest, up 11.
    Stars / 30 days
    +2
    npm / typical week
    Ships no npm package
    Tools exposed
    2
    Last commit
    31 days ago
    Commits / 12 weeks
    14
    Weekly commits over the last 12 weeks, oldest first: 0, 0, 0, 0, 4, 0, 4, 6, 0, 0, 0, 0. 14 in total, and the last figure is the current week so far.
    Maintenance grade
    B
    Tool descriptions
    A
    Score 53.0: show every number behind it
    • Adoption56 / 100 · weight 40%
      • GitHub stars56
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance92 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade80
      • Commit cadence85
    • Momentum29 / 100 · weight 14%
      • Stars gained, relative to size35
      • Stars gained, absolute21
      • npm download trendnot measured
        no npm download history
    • Tool quality81 / 100 · weight 13%
      • Tool description quality73
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    67.9
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    53.0
  • 17

    MinerU Document Explorer

    Abandoned but popular▼ 1
    opendatalab/MinerU-Document-Explorer ↗

    Best for: Building a literature knowledge base from a folder of PDFs: its Retrieve, Deep Read, and Ingest suites let an agent search, navigate, and ingest documents.

    The server exposes MCP tool suites for Retrieve, Deep Read, and Ingest, covering cross-collection search, section-level document navigation, and ingestion into an LLM wiki from Markdown, PDF, DOCX, and PPTX. Before choosing it, note the dependency: it runs through the qmd CLI and, in MCP mode, keeps embedding and reranking models loaded in memory across requests.

    GitHub stars
    616
    Stars over the last 90 days: 585 at the earliest of 9 readings and 627 at the latest, up 42.
    Stars / 30 days
    +17
    npm / typical week
    49
    Tools exposed
    never inspected
    Last commit
    130 days ago
    Commits / 12 weeks
    0
    Maintenance grade
    D
    Tool descriptions
    Not graded
    Score 51.0: show every number behind it
    • Adoption72 / 100 · weight 40%
      • GitHub stars70
      • npm downloads18
        downloads show none of the weekday rhythm human traffic has; halved
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance46 / 100 · weight 24%
      • Last commit touching this server68
      • Repository maintenance grade30
      • Commit cadence5
    • Momentum50 / 100 · weight 14%
      • Stars gained, relative to size61
      • Stars gained, absolute51
      • npm download trend29
    • Tool quality≈73 / 100 · weight 13%
      • Tool description quality≈73
        tool descriptions not yet scored
      • Built and inspected by Glamanot measured
        never built and inspected by Glama
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integrates100
    Weighted mean of the five
    65.4
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: widely adopted
    1.00
    Composite score
    51.0
  • 18

    opentakeoff-mcp

    Emerging▼ 1
    Kentucky-ai/opentakeoff ↗

    Best for: For agent-driven construction takeoffs from plan PDFs: 42 MCP tools cover scaling, room measurement, and DXF/PDF exports.

    It runs locally over stdio via npx -y opentakeoff-mcp and exposes 42 tools covering plan PDF loading, scale setting and verification, one-click room detection, length and area measurement, and DXF and marked-PDF export. Before choosing it, know that it assumes plans are available as local PDFs and exchanges native takeoff data in OpenTakeoff's own annotation and report schemas.

    GitHub stars
    101
    Stars over the last 90 days: 19 at the earliest of 31 readings and 113 at the latest, up 94.
    Stars / 30 days
    +46
    npm / typical week
    Ships no npm package
    Tools exposed
    42
    Last commit
    yesterday
    Commits / 12 weeks
    561
    Weekly commits over the last 12 weeks, oldest first: 15, 8, 14, 202, 96, 48, 40, 31, 31, 42, 34, 27. 588 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    A
    Score 50.4: show every number behind it
    • Adoption50 / 100 · weight 40%
      • GitHub stars50
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance100 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence100
    • Momentum87 / 100 · weight 14%
      • Stars gained, relative to size100
      • Stars gained, absolute67
      • npm download trendnot measured
        no npm download history
    • Tool quality83 / 100 · weight 13%
      • Tool description quality75
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integratesnot measured
        not published by the vendor it integrates
    Weighted mean of the five
    76.0
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: modest but real audience
    0.85
    Composite score
    50.4
  • 19

    KDAN PDF

    SteadyNo change
    KDAN-PDF/KDAN-PDF-MCP ↗

    Best for: For chat-based PDF cleanup before sharing: a remote MCP connector for compressing, removing pages, redacting, protecting, and comparing PDFs in Claude or ChatGPT.

    KDAN PDF is a remote MCP server added to Claude or ChatGPT by pasting its connector URL; its README documents natural-language workflows for compressing, deleting pages, redacting, protecting, and comparing PDFs. The exact tool list was not independently verified and no npm package is published, so the live server's tool surface remains to be confirmed.

    GitHub stars
    57
    Stars / 30 days
    0
    npm / typical week
    Ships no npm package
    Tools exposed
    never inspected
    Last commit
    13 days ago
    Commits / 12 weeks
    5
    Weekly commits over the last 12 weeks, oldest first: 0, 0, 0, 0, 0, 0, 3, 0, 0, 1, 0, 0. 4 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    Not graded
    Score 49.6: show every number behind it
    • Adoption44 / 100 · weight 40%
      • GitHub stars44
      • npm downloadsnot measured
        no npm package
      • Used through Glamanot measured
        not used through Glama in the last 30 days
    • Maintenance93 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence65
    • Momentum0 / 100 · weight 14%
      • Stars gained, relative to size0
      • Stars gained, absolute0
      • npm download trendnot measured
        no npm download history
    • Tool quality≈73 / 100 · weight 13%
      • Tool description quality≈73
        tool descriptions not yet scored
      • Built and inspected by Glamanot measured
        never built and inspected by Glama
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integrates100
    Weighted mean of the five
    58.4
    × relevance: the keyword is dedicated here
    1.00
    × continuity: actively changing
    1.00
    × evidence: modest but real audience
    0.85
    Composite score
    49.6
  • 20

    EnConvert MCP Server

    Emerging▼ 2
    enconvert/mcp ↗

    Best for: Teams needing file conversion plus live-web data work: it exposes conversion, rendering, search, extraction, ingest, and page-monitoring tools.

    The server exposes 24 tools for file conversion, image compression, Markdown/PDF generation, URL perception, web search, structured extraction, RAG ingest, and page-change monitoring. The web-data tools require a private API key and are plan-gated, so assess subscription access before relying on them.

    GitHub stars
    7
    Stars / 30 days
    no snapshot history
    npm / typical week
    98
    Tools exposed
    24
    Last commit
    24 days ago
    Commits / 12 weeks
    15
    Weekly commits over the last 12 weeks, oldest first: 0, 0, 0, 0, 4, 1, 1, 4, 5, 0, 0, 0. 15 in total, and the last figure is the current week so far.
    Maintenance grade
    A
    Tool descriptions
    A
    Score 48.6: show every number behind it
    • Adoption43 / 100 · weight 40%
      • GitHub stars23
      • npm downloads21
        downloads show none of the weekday rhythm human traffic has; halved
      • Used through Glama40
    • Maintenance97 / 100 · weight 24%
      • Last commit touching this server100
      • Repository maintenance grade100
      • Commit cadence85
    • Momentum82 / 100 · weight 14%
      • Stars gained, relative to sizenot measured
        no snapshot history yet
      • Stars gained, absolutenot measured
        no snapshot history yet
      • npm download trend82
    • Tool quality93 / 100 · weight 13%
      • Tool description quality85
      • Built and inspected by Glama100
    • Trust100 / 100 · weight 9%
      • License100
      • Published by the vendor it integrates100
    Weighted mean of the five
    73.3
    × relevance: the keyword is declared here
    0.78
    × continuity: actively changing
    1.00
    × evidence: modest but real audience
    0.85
    Composite score
    48.6

Questions people ask

Which pdf-mcp should I use for extracting tables and technical diagrams?

The pdf-mcp (I-CAN-hack/pdf-mcp) server is built for this: it exposes 5 tools and renders pages to images, which helps LLMs extract text, tables, and technical diagrams. The pdf-mcp (jztan/pdf-mcp) server has more tools, 13, but does not mention page rendering, so for diagram-heavy PDFs the I-CAN-hack one is the better choice.

Is Markdownify MCP Server still maintained?

No, it is marked 'Abandoned but popular' and shows no recent activity: last commit 121 days ago and 0 commits in the last 12 weeks. Use it only for quick experiments, not for production.

What does Docling MCP require to run?

It is intended for teams already running Docling Serve, because its remote conversion mode exposes the conversion tools. If you do not have Docling Serve, you would rely on local fallback, but the description emphasizes the remote conversion mode.

Can kordoc handle regular English PDFs?

kordoc parses PDF into Markdown, but its focus is South Korean document formats like HWP, HWPX, and PDF, and it is optimized for administrative and public institution files. For general English PDFs you might prefer a server without that specialization, such as pdf-mcp (jztan/pdf-mcp).

Other comparisons