benspdf
<!-- mcp-name: io.github.benbergner/benspdf -->
<div align="center">
<h1>BensPDF</h1>
<p><strong>PDF tools for AI agents. Your files never leave your machine.</strong></p>
<p>
<a href="https://pypi.org/project/benspdf-mcp/"><img src="https://img.shields.io/pypi/v/benspdf-mcp?style=flat-square&logo=pypi&logoColor=white&label=PyPI" alt="PyPI"></a>
<a href="https://pypi.org/project/benspdf-mcp/"><img src="https://img.shields.io/pypi/pyversions/benspdf-mcp?style=flat-square&logo=python&logoColor=white" alt="Python versions"></a>
<a href="https://github.com/benbergner/BensPDF/blob/main/LICENSE"><img src="https://img.shields.io/pypi/l/benspdf-mcp?style=flat-square" alt="License"></a>
<a href="https://registry.modelcontextprotocol.io/v0/servers?search=benspdf"><img src="https://img.shields.io/badge/MCP-registry-1f6feb?style=flat-square" alt="MCP registry"></a>
</p>
<p><strong>Install in one click</strong></p>
<p>
<a href="https://vscode.dev/redirect/mcp/install?name=benspdf&config=%7B%22type%22%3A%22stdio%22%2C%22command%22%3A%22uvx%22%2C%22args%22%3A%5B%22benspdf-mcp%22%5D%7D"><img src="https://img.shields.io/badge/VS_Code-Install-0098FF?style=for-the-badge&logo=visualstudiocode&logoColor=white" alt="Install in VS Code"></a>
<a href="https://cursor.com/en/install-mcp?name=benspdf&config=eyJuYW1lIjoiYmVuc3BkZiIsInR5cGUiOiJzdGRpbyIsImNvbW1hbmQiOiJ1dngiLCJhcmdzIjpbImJlbnNwZGYtbWNwIl19"><img src="https://img.shields.io/badge/Cursor-Install-000000?style=for-the-badge&logo=cursor&logoColor=white" alt="Install in Cursor"></a>
</p>
</div>
---
Ben's PDF tools for AI agents, exposed over the [Model Context Protocol](https://modelcontextprotocol.io) (MCP).
> [!NOTE]
> Your PDFs are read on your own machine and never uploaded. With a hosted model
> your questions still reach that model; pair the tools with a local Ollama model
> and nothing leaves the machine at all.
Works with Claude Desktop, Claude Code, VS Code, Kiro, Cursor, the ChatGPT
desktop app, and any other MCP client. The one-click buttons above need
[uv](https://docs.astral.sh/uv/getting-started/installation/); for every other
client, or to do it by hand, see [Setup](#setup).
## Tools
| Tool | What it does |
| --- | --- |
| `pdf_page_count` | Counts the pages in a PDF |
| `pdf_metadata` | Reads document properties: title, author, dates, producer |
| `pdf_check_text` | Says whether a PDF is readable text or a scan that needs OCR |
| `pdf_page_layout` | Page sizes, orientation, rotation and page boxes |
| `pdf_check_access` | Encryption, and what the file permits: printing, copying, editing |
| `pdf_render_pages` | Renders pages to images, so a page can be looked at |
| `create_test_pdf_file` | Generates a throwaway PDF, handy for trying things out |
| `export` | Saves results to a real location on disk |
| `list_artifacts` | Lists recent temporary results |
| `discard` | Deletes temporary results now |
## Where results go
When a tool makes a new PDF, it goes into a scratch folder instead of your own
folders, and you get back a short id like `art_a1b2c3d4.pdf`. Tools accept those
ids anywhere they accept a file path, so several steps can be chained together.
Results carry the artifact's `path` as well as its id, so you can open a rendered
page or an intermediate file straight away without exporting it first.
`export` is the only tool that writes into your folders, so nothing shows up
until you ask for it. Each time the server starts it clears out scratch files
older than 7 days. Set `BENSTOOLS_WORKSPACE` to put the scratch folder somewhere
other than `~/.benstools/work`.
## Setup
Install [uv](https://docs.astral.sh/uv/getting-started/installation/)
once, then point your client at `uvx benspdf-mcp` and uv fetches the package,
plus a suitable Python, on first run.
```bash
# macOS and Linux
curl -LsSf https://astral.sh/uv/install.sh | sh
# Windows
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
```
On macOS, `brew install uv` works too.
Add the server to your client with one of the configs below, then restart the
server from your client's UI. Once connected, just ask in plain language:
> How many pages are in ~/Downloads/report.pdf?
### Claude Desktop
Edit `claude_desktop_config.json`, which lives at
`~/Library/Application Support/Claude/` on macOS and `%APPDATA%\Claude\` on
Windows. You can also open it from **Settings → Developer → Edit Config**.
```json
{
"mcpServers": {
"benspdf": {
"command": "uvx",
"args": ["benspdf-mcp"]
}
}
}
```
### Claude Code
One command, no config file. Add `--scope user` to enable it everywhere rather
than just the current project.
```bash
claude mcp add benspdf -- uvx benspdf-mcp
```
Check it connected with `claude mcp list`.
### VS Code
`.vscode/mcp.json` in your workspace, or the same file in your user profile.
Note VS Code uses `servers` rather than `mcpServers`, and wants an explicit
`type`.
```json
{
"servers": {
"benspdf": {
"type": "stdio",
"command": "uvx",
"args": ["benspdf-mcp"]
}
}
}
```
### Kiro
`.kiro/settings/mcp.json` in your workspace, or `~/.kiro/settings/mcp.json` to
enable it everywhere. `autoApprove` skips the confirmation prompt for tools you
trust.
```json
{
"mcpServers": {
"benspdf": {
"command": "uvx",
"args": ["benspdf-mcp"],
"autoApprove": ["pdf_page_count"]
}
}
}
```
### Cursor
The button at the top does this for you. By hand, it's `~/.cursor/mcp.json` to
enable it everywhere, or `.cursor/mcp.json` in a project.
```json
{
"mcpServers": {
"benspdf": {
"type": "stdio",
"command": "uvx",
"args": ["benspdf-mcp"]
}
}
}
```
### Windsurf
`~/.codeium/windsurf/mcp_config.json`. You can also reach it from Cascade:
**Settings → Cascade → Manage MCPs → View raw config**, which is worth using
since the path has moved between versions.
```json
{
"mcpServers": {
"benspdf": {
"command": "uvx",
"args": ["benspdf-mcp"]
}
}
}
```
### Continue
`~/.continue/config.yaml`, or a `.yaml` file under `.continue/mcpServers/` in a
project. Continue is the one client here that doesn't take the JSON shape above:
its config is YAML, and `mcpServers` is a list rather than an object keyed by name.
```yaml
mcpServers:
- name: benspdf
command: uvx
args:
- benspdf-mcp
```
### ChatGPT desktop app, Codex CLI, Codex IDE extension
All three are Codex clients and share one config file, `~/.codex/config.toml`,
so adding the server once covers all of them. Note this one is TOML, not JSON.
```toml
[mcp_servers.benspdf]
command = "uvx"
args = ["benspdf-mcp"]
```
### Any other MCP client
Almost every client uses the same JSON as Claude Desktop above — an `mcpServers`
object, with `command` set to `uvx` and `args` to `["benspdf-mcp"]`. Some want an
explicit `"type": "stdio"`; adding it is harmless where it isn't required.
If a client just asks for a command to run, it's:
```
uvx benspdf-mcp
```
## Fully offline with Ollama
The clients above keep your PDFs local, but they answer using a hosted model.
Pair the tools with a local model instead and nothing leaves your machine.
You'll need [Ollama](https://ollama.com) with a model pulled, plus this package:
```bash
pip install benspdf-mcp ollama
ollama pull llama3.1
```
Then run the bundled CLI:
```bash
benspdf-cli # uses the first model you have
benspdf-cli --model llama3.1 # or pick one
BENSPDF_MODEL=llama3.1 benspdf-cli # or set it once
```
`python -m benspdf.cli` does the same thing, handy from a source checkout.
```
You: how many pages in ~/Downloads/report.pdf?
[Using tool: pdf_page_count]
[Result: 12 pages in report.pdf]
Assistant: The PDF has 12 pages.
```
## Reference
Each tool's own description tells your client what it does and when to use it, so
in normal use there is nothing to look up. If you want the detail — every field a
tool returns, and the reasoning behind the answers it gives — see the
[tool reference](https://github.com/benbergner/BensPDF/blob/main/docs/tools.md).
To work on the code, see
[CONTRIBUTING.md](https://github.com/benbergner/BensPDF/blob/main/CONTRIBUTING.md).
## License
Apache 2.0
TDQS
Scored across 10 tools
Each of the six PDF inspection tools targets a genuinely different aspect: page count, metadata stores, text layer/OCR need, encryption/permissions, page geometry, and visual rendering. Descriptions actively cross-reference and delimit each other (e.g. 'for whether pages hold readable text use pdf_check_text'), leaving no real overlap. The workspace tools (discard/list_artifacts/export/create_test_pdf_file) are likewise distinct in purpose.
Most PDF tools share a 'pdf_' prefix, but verb/noun ordering is inconsistent (pdf_page_count and pdf_metadata are noun-first while pdf_check_text and pdf_render_pages are verb-first), and the four workspace tools drop the prefix entirely (discard, export, list_artifacts, create_test_pdf_file). All snake_case keeps it readable, but the convention is mixed rather than predictable.
Ten tools is well-scoped for a PDF utility server, with a coherent split between PDF inspection (six tools), artifact/file lifecycle (three tools), and a test-fixture generator. Each tool earns its place without redundancy.
Several descriptions explicitly point users to pdf_extract_text and pdf_ocr ('use pdf_extract_text', 'prefer pdf_ocr'), yet neither tool is in the set, so the surface can check for text and render pages but cannot extract text or run OCR. Core inspection and output/export flows are covered, but this text-extraction hole is a notable gap.