pdf-modifier-mcp

PDF Modifier MCP

Give your AI coding assistant the ability to natively read, edit, and redact PDF documents.

Most AI assistants can generate Python scripts to edit PDFs, but running them requires manual execution, debugging missing fonts, and fixing broken layouts. PDF Modifier is an MCP server that gives your AI direct, native tools to safely manipulate PDFs in place.

The numbers:

Metric	Without PDF Modifier MCP	With PDF Modifier MCP
Workflow	AI writes script -> You run it -> Debug errors	AI edits PDF directly via MCP
Time to redact 10 invoices	~15 minutes (trial & error)	Seconds (autonomous)
Layout preservation	Often breaks styling or coordinates	100% (Base 14 font matching)
Interface	Terminal only	Native AI Tools + CLI fallback

Quick Install (30 seconds)

One command. No cloning, no venv. Use user scope (-s user) so the tools are available across all your projects.

Claude Code

claude mcp add -s user pdf-modifier -- uvx --upgrade pdf-modifier-mcp

Gemini CLI

gemini mcp add -s user pdf-modifier uvx -- --upgrade pdf-modifier-mcp

OpenAI Codex CLI

Add to ~/.codex/config.toml:

[mcp_servers.pdf-modifier]
command = "uvx"
args = ["--upgrade", "pdf-modifier-mcp"]

GitHub Copilot (VS Code)

Add to .vscode/mcp.json or your User Settings:

{
  "mcp": {
    "servers": {
      "pdf-modifier": {
        "command": "uvx",
        "args": ["--upgrade", "pdf-modifier-mcp"]
      }
    }
  }
}

Then ask your assistant:

"Read the structure of invoice.pdf and redact all credit card numbers"

That's it. You're running.

What You Get

5 MCP Tools — PDF manipulation, on demand

Tool	What it does
read_pdf_structure	Extract complete PDF structure with text positions, sizes, and fonts
inspect_pdf_fonts	Search for terms and report their exact font properties
list_pdf_hyperlinks	Inventory all existing hyperlinks and URIs in the document
modify_pdf_content	Find and replace text (or regex) with strict style and layout preservation
batch_modify_pdf_content	Apply the same replacements to multiple PDFs at once

CLI Interface (For Humans)

Still want to script it yourself? The package includes a powerful Typer CLI:

# Simple text replacement
pdf-mod modify input.pdf output.pdf -r "old text=new text"

# Inventory hyperlinks
pdf-mod links input.pdf

# Batch process multiple PDFs
pdf-mod batch *.pdf -o output/ -r "Draft=Final"

Regex replacement (dates, IDs, etc.)

pdf-mod modify input.pdf output.pdf -r "Order #\d+=Order #REDACTED" --regex

Create hyperlinks

pdf-mod modify input.pdf output.pdf -r "Click Here=Visit Site|https://example.com"

## Before / After

**Before PDF Modifier MCP** — Manual scripting workflow:
```python
# The AI generates a script like this for you to run:
import fitz
doc = fitz.open("invoice.pdf")
for page in doc:
    # Hope the AI calculated the rects and fonts correctly...
    page.insert_text((100, 100), "REDACTED", fontname="helv")
doc.save("out.pdf")

With PDF Modifier MCP — Autonomous AI action:

# The AI natively calls the tool without asking you to run scripts
modify_pdf_content(
    input_path="invoice.pdf",
    replacements={"pattern": r"\b\d{4}-\d{4}-\d{4}-\d{4}\b", "replacement": "XXXX-XXXX-XXXX-XXXX"},
    use_regex=True
)

Architecture

MCP Host (Claude, Gemini, Copilot)
  ↓ MCP protocol (stdio)
pdf-modifier-mcp (FastMCP Server)
  │
  ├── Core Layer (PDFModifier, PDFAnalyzer)
  │     ├── Regex matching & text extraction
  │     ├── Base 14 font fallback mapping
  │     └── Layout & coordinate preservation
  │
  └── PyMuPDF (fitz) Engine

Development

See CONTRIBUTING.md for setup, code standards, and PR workflow.

git clone https://github.com/mlorentedev/pdf-modifier-mcp.git
cd pdf-modifier-mcp
make setup     # create venv + install deps
make check     # lint + typecheck + test

License

MIT License - see LICENSE for details.

Documentation

Project-bound knowledge lives in docs/ (docs-as-code): ADRs, runbooks, troubleshooting, and lessons.