Stirling PDF

What it is

Stirling PDF is an editor and processor for PDF files, made by Stirling PDF Inc. Its README says you can run it as a desktop app, in a browser or on your own servers with a private API. The tools merge, split, compress, convert, OCR, sign and redact, and a saved chain of tools can run on any files. The README promises that you can edit, sign, redact, convert and automate PDFs “without sending documents to external services”.

1. You install the desktop app or start the Docker image
2. You open the app, or http://localhost:8080 for a server
3. You pick a tool, add your PDFs and run it
4. Local Only mode and your own server keep files with you
5. Scripts and AI agents call the same tools by API or MCP

Who it suits

Anyone who edits, merges, compresses, signs or redacts PDFs and would rather do it on their own machine. IT teams that host one tool for a whole company fit it, and so do developers who call PDF tools from scripts or agents. Read its license before a company runs it in production.

Where to use it

Stirling PDF runs as a desktop app on Windows, macOS and Linux, as a server in Docker, Java or Kubernetes, and as Stirling Cloud in a browser. These are the ways to connect other tools.

• Claude Code through .mcp.json, and VS Code through .vscode/mcp.json
• Codex CLI through ~/.codex/config.toml
• Claude Desktop through the mcp-remote bridge, which needs Node.js
• Your own scripts, through the REST API and the Pipeline API
• n8n, Zapier, Make and Power Automate, through their generic HTTP request action
• Telegram, through a bot on a self-hosted server
• Watched folders on a server, which run a saved workflow on every file dropped in

How to set it up

Get it saves a zip of Mindket’s setup notes in your downloads folder, with a page for each way to run it from the maker’s README and docs. The shortest routes are the desktop installer and one Docker line.

1. Pick a route: the desktop app for yourself, or Docker for a server.
2. For the desktop app, install it with the installer, winget or Homebrew, then open Stirling PDF. On the first screen choose Continue without signing in to work in Local Only mode.
3. For a server, run the line below and open http://localhost:8080.
4. Sign in as admin with the password stirling, then set a new password when asked.

docker run -p 8080:8080 docker.stirlingpdf.com/stirlingtools/stirling-pdf

winget install StirlingTools.StirlingPDF

The first line is the README’s quick start. It mounts no volume, so use the longer Docker command in the notes for anything you keep. The second line installs the desktop app on Windows.

How to use it

Open the tool picker, choose a tool, add your PDFs and run it. The Tools page puts the most used at the top: PDF Text Editor, Page Editor, Merge, Compare, Compress, Convert, OCR and Redact. In the desktop app, Ctrl+S saves over the file you opened and Save As makes a copy. On a server, every server-side tool also has an endpoint. A few tools, such as the viewer and drawing a signature, run only in the browser and have none. The REST API page compresses a file like this:

curl -X POST "$STIRLING_URL/api/v1/misc/compress-pdf" \
  -H "X-API-KEY: $STIRLING_API_KEY" \
  -F "fileInput=@document.pdf" \
  -F "optimizeLevel=5" \
  -o compressed.pdf

Set STIRLING_URL to your server’s address and STIRLING_API_KEY to a key from Settings, then Preferences, then API Keys.

Example

You have a PDF that needs repairing and shrinking, and you want one script to do both. The Pipeline API page sends the two steps in one request.

curl -X POST "$STIRLING_URL/api/v1/pipeline/handleData" \
  -H "X-API-KEY: $STIRLING_API_KEY" \
  -F "fileInput=@input.pdf" \
  -F 'json={
    "name": "Repair then compress",
    "pipeline": [
      {"operation": "/api/v1/misc/repair", "parameters": {}},
      {"operation": "/api/v1/misc/compress-pdf", "parameters": {"optimizeLevel": 2}}
    ]
  }' \
  -o result.pdf

One output file comes back as the file itself, here result.pdf. Several files would come back as output.zip.

For newcomers and experts

If you are new. Start with the desktop app. On the first screen choose Continue without signing in, which gives Local Only mode, then open a PDF and try Merge or Compress. A tool with a cloud badge needs Stirling Cloud or a server, and running one asks you to sign in or connect.

For experts. Every key in settings.yml also works as an environment variable, such as SECURITY_ENABLELOGIN or SYSTEM_DEFAULTLOCALE. Pipeline steps name the full endpoint path, and ?async=true on a tool call returns a job ID to poll. A server lists every endpoint at /swagger-ui/index.html, with the OpenAPI spec at /v1/api-docs.

Cluster mode runs several pods behind one service. The Kubernetes page says it needs a Team or Enterprise license plus your own PostgreSQL, Valkey and S3. The MCP server takes OAuth 2.1 or per-user API keys, and an admin can list the tools it allows or blocks.

Good to know

Most of it runs on your own machine or server, and a few paths reach out.

• License: the README says Stirling PDF is open-core. The root LICENSE is MIT, “Copyright (c) 2025 Stirling PDF Inc.”, for content outside the folders it lists. Those folders have their own LICENSE files, and the four we read hold the Stirling PDF User License, which allows production use only with a valid user license. GitHub shows “Other”. Read both licenses before a company runs it
• Accounts: none for the desktop app in Local Only mode or for a server you run. Stirling Cloud lets guests use the manual tools, but API keys, MCP and AI tools need an account
• What leaves your computer: in Local Only mode every tool runs on your computer and nothing is uploaded. In Stirling Cloud mode, tools the built-in engine cannot run go to Stirling Cloud. The Cloud docs say its AI features use Anthropic models and Voyage AI embeddings, and see excerpts, never whole files
• Analytics: a server can send anonymous usage data to PostHog’s EU servers, and browsers can load a Scarf pixel. The docs say document contents are never sent. The first admin is asked on first sign-in, and SYSTEM_ENABLEANALYTICS=false turns it all off
• First sign-in: the account admin with the password stirling is made on an empty database. With login off, anyone who can reach the server can use it
• Deleting: Redact removes text for good, so keep a copy of the original. Folder scanning deletes the input files after a successful run
• Plans: a server runs on the Free plan (5 users) until you add Team or Enterprise. Processor automation, AI tools and PDF tool calls made with an API key draw on a monthly processing allowance. A Team licence key stops API-key calls counting, and Enterprise counts none of it. When it runs out, API calls return 402 with ACCOUNT_LINK_REQUIRED
• Telegram bot: the docs call it a community-built beta, not covered by Team or Enterprise support
• Install lines: the Proxmox one-liner in the Docker notes runs a script from the internet, so read it first. On Windows x64 the desktop app installs for all users, so its update prompt may need administrator permissions
• Updates: the desktop app checks for a new version shortly after it starts, and asks before installing by default. Pin a version tag for Docker instead of latest