Skip to main content

​
Week of May 22, 2026 — Launch week

PDFBase is live. Here’s everything that shipped this week.

​
New features

  • Free web tools — Six no-signup tools for quick conversions and extraction: HTML to PDF, URL to PDF, Markdown to PDF, PDF text extractor, PDF table extractor, and invoice PDF generator. Use these to try the platform before integrating the API.
  • TypeScript and Python SDKs — Official SDKs with retries, typed errors, and full API coverage. Get started with the TypeScript SDK or Python SDK.
  • Command-line interface — Run conversions, merges, splits, and extractions from your terminal. See the CLI reference.
  • MCP server — Connect PDFBase to Claude, Cursor, and other AI agents through nine MCP tools wrapping the REST API. See the MCP overview.
  • PDF hosting — Host generated PDFs at a permanent public URL with no expiry. Useful for invoices, receipts, and reports you want to share by link.
  • PDF compression — New POST /v1/pdfs/compress endpoint strips metadata and optimizes object streams. Response includes the savings percentage. See compress.
  • Real OCR and AI-structured extraction — OCR is now powered by Tesseract for true text recognition on scanned PDFs. Structured extraction uses Claude to turn documents into JSON with built-in presets. See OCR and structured extraction.
  • Batch processing — Submit many jobs in one request with parallel execution and progress tracking. See batch processing.
  • Webhooks — HMAC-SHA256 signed delivery with 5-attempt retry. See webhooks.
  • Dashboard — Manage API keys, view usage, and check credit balance at pdfbase.dev/dashboard.

​
Updates

  • Idempotency keys — All mutating endpoints accept an Idempotency-Key header with 24-hour replay protection. See idempotency.
  • Rate limiting — Token-bucket rate limits with clear response headers. See rate limits.
  • Credit balance headers — Every response now includes your current credit balance, and requests return a clean 402 when you’re out. See pricing.
  • Render concurrency limit — A new per-account limit of three concurrent Chromium renders keeps latency stable under load.
  • Hardened security — SSRF protection on edit, convert, and extract routes; sanitized headers on hosted PDFs; and authenticated /metrics access. See security.
  • Better template engine — Handlebars templates now validate against a schema and ship with built-in helpers. See templates.
  • Structured logging and request IDs — Every request is tagged with an ID for easier debugging when contacting support.

​
Bug fixes

  • Containers now bind to 0.0.0.0 so deployments behind Fly.io and similar platforms route traffic correctly.
  • Switched to puppeteer-core with system Chromium to fix render failures in container environments.
  • Improved error logging when Chromium fails to launch.
Follow @pdfbase for release announcements.