Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

Stirling PDF

Self-hosted PDF editor and API with more than 50 tools, plus an optional AI engine and MCP server under a separate commercial licence.

Reviewed by ReviewRamble editors · Updated

View on GitHub Free plan

Share this page:XLinkedInFacebookRedditWhatsAppEmail

Our rating

4.0/5

The self-hosted PDF toolbox to beat, now with AI and MCP on top, provided you read the licence before switching those parts on.

What is Stirling PDF?

Stirling PDF is a PDF platform you run as a desktop app, in the browser or on your own server with a private REST API. It edits, merges, splits, signs, redacts, converts, compresses and OCRs PDFs, chains tools into no-code pipelines, and supports SSO. One Docker command starts it on port 8080.

AI is off by default. An admin can turn on six capabilities: a chat assistant, document questions, PDF creation from a prompt, a maths auditor, a PDF comment agent and document classification. They run on your own AI engine with Anthropic, OpenAI, Ollama or any OpenAI-compatible model, or on Stirling Cloud AI. A built-in MCP server, also off by default, lets Claude, Codex and other clients run Stirling tools at /mcp.

The project is open core: code outside a set of directories is MIT, while the AI engine and the proprietary app module, which holds the MCP server, need a paid user licence for production use. The repo has about 96,758 GitHub stars and 9,367 forks; v3.3.0 shipped on 5 November 2026.

Is Stirling PDF worth it?

Stirling PDF earned its stars as a private replacement for the PDF websites people upload contracts to. Merging, redacting, signing and OCR run on your own server, and nearly every tool has a REST endpoint, so it slots into a pipeline as easily as into a browser tab.

The AI layer is newer and well fenced. It stays off until an admin enables it, each of the six capabilities can be turned off on its own, and the docs spell out where content goes: language model providers receive prompts, relevant document text, conversation history and file names, and embedding providers receive document text. With Ollama for both models and embeddings, nothing leaves your network. The defaults point elsewhere, though: Anthropic's claude-haiku-4-5 for language and VoyageAI's voyage-4 for embeddings, so a quick setup sends documents to two outside companies.

The licence is the part to read twice. The MIT badge covers the core tools, but the AI engine and the module holding the MCP server sit under a commercial licence that forbids production use without a subscription. Without Enterprise, AI calls also draw on the monthly processing allowance even when you supply your own engine and key. The pricing page itself is muddled, quoting 1,000 monthly credits on the Free plan and the first 500 runs free elsewhere.

Our take: as a PDF server it is the default choice for a self-hosted stack, and the MCP server is a tidy way to let an agent fill, redact or convert files. The AI features are a sensible extra for teams already paying for Team or Enterprise, not a reason to adopt it on their own.

Where it earns its place

  • A 30-person law firm

    Redact and merge client files without uploading them to a website

    Run the Docker image on an office server behind SSO and keep AI off. The core tools are MIT, and the Free plan covers 5 users; past that, Team is $99/month for up to 100.

  • A platform engineer at a logistics company

    Let a coding agent convert, compress and fill PDFs inside a workflow

    Enable the MCP server with OAuth 2.1 or per-user API keys and point Claude Code or Codex at https://your-server/mcp. Every call runs and is audited as a Stirling user; production use of this module needs a licence.

  • A compliance team in a regulated bank

    Ask questions of policy PDFs without data leaving the network

    Run your own AI engine with Ollama for the language model and embeddings, keep port 5001 private and set a shared secret. Enterprise removes the processing allowance on AI calls; price on request.

Before you commit

Setup
docker run -p 8080:8080 docker.stirlingpdf.com/stirlingtools/stirling-pdf, then open localhost:8080. Desktop, JAR and Kubernetes installs are documented. AI needs an admin to pick a mode under Settings, Server, AI Engine.
Integrations
REST API for nearly all tools; MCP server for Claude, Codex and VS Code; AI providers Anthropic, OpenAI, Ollama and OpenAI-compatible endpoints; VoyageAI, OpenAI or Ollama embeddings; OAuth2/OIDC SSO, SAML2 on Enterprise.
Data and privacy
Self-hosted, with nothing leaving the server while AI is off. Turned on, prompts and document text go to the providers you configure, or to Stirling Cloud, which keeps indexed text only if you enable that. The engine on port 5001 carries provider keys and must stay private.
Cost at scale
Team is $99/month per 100-user block. Processing beyond the allowance costs 1¢ per credit; Enterprise, priced on request, lifts the allowance for AI and adds air-gapped deployment and audit logs.
Limits
462 open issues and pull requests; Office Convert is still beta and set to replace LibreOffice; the latest-ultra-lite image has no plans; free servers cap at 5 users.

What’s good

  • More than 50 PDF tools with a REST API, self-hosted
  • AI and MCP both off by default, with clear data-flow docs
  • Local models through Ollama keep documents on your network
  • Team plan is a flat $99/month for 100 users

What’s not

  • AI engine and MCP module need a commercial licence in production
  • Default AI setup sends documents to Anthropic and VoyageAI
  • AI use counts against the processing allowance below Enterprise
  • Pricing page contradicts itself on free credits

Best for

Teams that handle sensitive PDFs and want editing, OCR and an agent-callable API on their own servers.

Stirling PDF alternatives

Stirling PDF: common questions

Is Stirling PDF worth it?
The self-hosted PDF toolbox to beat, now with AI and MCP on top, provided you read the licence before switching those parts on. We rate it 4.0 out of 5.
How much does Stirling PDF cost?
Free: $0 with 1,000 monthly credits and up to 5 server users. Team: $99/month for 100 users, more in blocks of 100. Enterprise: custom. Extra processing 1¢ per credit.
What are the downsides of Stirling PDF?
AI engine and MCP module need a commercial licence in production; default AI setup sends documents to Anthropic and VoyageAI; aI use counts against the processing allowance below Enterprise; pricing page contradicts itself on free credits.
Who is Stirling PDF best for?
Teams that handle sensitive PDFs and want editing, OCR and an agent-callable API on their own servers.
How much does Stirling PDF cost for a team or in production?
Team is $99/month per 100-user block. Processing beyond the allowance costs 1¢ per credit; Enterprise, priced on request, lifts the allowance for AI and adds air-gapped deployment and audit logs.
Is Stirling PDF safe for company data?
Self-hosted, with nothing leaving the server while AI is off. Turned on, prompts and document text go to the providers you configure, or to Stirling Cloud, which keeps indexed text only if you enable that. The engine on port 5001 carries provider keys and must stay private.
What are the best Stirling PDF alternatives?
The closest alternatives we have reviewed are Docling, Marker and MinerU.

Answers checked Nov 16, 2026.

Make Stirling PDF? Show this rating on your siteStirling PDF rated 4.0 out of 5 on ReviewRamble