Stripe webhook fires on payment success. Handler calls PDFops fill-form with the charge details, gets an invoice PDF back, stores it in R2 — all inside one Cloudflare Worker, no queue, no separate render service. Full handler code →

Invoice PDFs from Stripe webhooks

Generate and store an invoice PDF the moment a Stripe payment succeeds — one Cloudflare Worker.

pdfops.dev

If your AI agent outputs 'filled the form' but the PDF is blank, you've met flattened AcroForms. The Inspector shows whether fields survived — paste the PDF, read the truth. Free tool, works with any TypeScript or Python agent stack. #webdev

PDF field Inspector

See what fields a PDF actually has.

pdfops.dev

Hashable PDFs unlock a nice CI trick: snapshot-test your document output like you snapshot-test UI. If the bytes changed, the diff is real — no flaky timestamps. Only works when generation is deterministic, which is the whole point. #webdev

Snapshot-testing PDFs in CI

Byte-identical output makes PDFs diffable.

pdfops.dev

Node.js PDF filling without node-gyp, without puppeteer, without a 200MB layer: one fetch, a {field: value} map, deterministic bytes back. Works identically in Bun and Deno. Code → #webdev

Fill a PDF in Node.js

One fetch, no native deps.

pdfops.dev

Every PDF invoice/contract/report pipeline eventually hits the same wall: the doc renders fine locally, breaks in serverless. Root cause is almost always a native dep or a headless browser. We built the primitive that removes both. #serverless

PDFops

Deterministic PDF fill & merge — no native deps, no browser.

pdfops.dev

A PDF playground that shows you the exact fetch call for what you just did: fill fields, merge files, copy the TypeScript snippet, ship it to your Worker or Lambda. No signup for 100 req/month. #webdev

PDFops playground

Try fill + merge live, copy the exact API call.

pdfops.dev

Python PDF libs that need C extensions won't deploy to most serverless platforms. The workaround is boring and reliable: POST a JSON field map over HTTPS, get finished bytes back. Comparison with working code → #webdev

Fill a PDF in Python

Native libs vs one HTTP call, compared.

pdfops.dev

Week 3 of building pdfops in public: directory campaign at 55/301 submissions, first daily.dev direct post live, and the API is still free at 100 req/month. The constraint is distribution, not code. #buildinpublic

PDFops — deterministic PDF fill & merge

Edge-native PDF primitive for serverless and AI apps.

pdfops.dev

Serverless PDF generation trap: puppeteer-core + chrome-aws-lambda is 50MB+ of cold start to render a form. If the PDF already exists and you just need fields filled, that's one POST — from a Worker, Lambda, or Deno Deploy. #serverless

PDFops — PDF fill & merge for the edge

One HTTP call. No headless browser.

pdfops.dev

The most annoying part of PDF automation isn't writing values — it's discovering what the field names ARE. Paste any PDF into the Inspector and see every AcroForm field, typed. Free, no signup, works for any Node.js or Python stack. #webdev

PDF field Inspector

Paste a PDF, see its AcroForm fields instantly.

pdfops.dev

puppeteer in a Lambda is ~50MB of Chromium shipped to render one invoice, and the cold start alone can outrun your request budget. Half of 'serverless PDF' pain is just that you brought a whole browser to a string-templating fight.

flattening a filled PDF is the step everyone forgets. Setting the fields read-only isn't enough — a determined user just re-enables editing. You have to bake the field values into static page content so the form is gone, not just locked. Two different operations.

if your PDF endpoint spins up headless Chrome per request, your unit cost is dominated by cold-start CPU, not the PDF. That's why serverless PDF bills feel random — you're billed for browser boots, not documents. A tiny pure-JS runtime makes the cost boring and predictable.

hot take: most 'PDF generation' services are a headless browser behind an HTTP call, and you're paying for the browser boot, not the document. HTML-to-PDF is the wrong primitive for filling a form that already exists — you don't re-render a page to set one field.

'same input, same bytes' sounds trivial until you try it with a PDF. Creation date, mod date and a random file ID get stamped in by default, so two identical fills hash differently. Zero those three and you can cache, diff and sign PDFs in CI like any other build artifact.

the reason your filled PDF looks blank in Preview but fine in Chrome: you set the field value but not the appearance stream, and some viewers only render the cached appearance. NeedAppearances=true, or generate the stream yourself. Classic AcroForm footgun.

why PDF is so annoying on the edge: Workers, Deno and Bun have no Chromium and no filesystem-heavy native deps. Every mainstream PDF tool assumes one or the other. The libs that actually run there are pure-JS or WASM and almost nobody knows their names.

TIL you can fill most PDF forms without rendering anything. AcroForm fields are just named key/values in the file's object tree — set the value, flip the appearance flag, done. No browser, no layout engine. People reach for headless Chrome to do what's basically a dictionary update.

Browser JS can fill a PDF client-side; a server API fits when you need determinism or the file can't leave the backend. We compared both paths for TypeScript apps, with working code for each. #webdev

Fill a PDF in JavaScript

Browser vs server-side PDF filling, compared with code.

pdfops.dev

Why do PDF APIs return different bytes for the same input? Timestamps, object ordering, and random IDs. We wrote up why we made ours byte-identical — so Node.js and serverless apps can hash, cache, and diff PDFs in CI. #webdev

The case for deterministic PDF filling

Same input, byte-identical output — hash it, cache it, sign it.

pdfops.dev

We filled a PDF form inside a Cloudflare Worker — no Chromium, no Lambda fallback, one fetch call. The AcroForm write is the easy part once you drop the headless-browser assumption. Full write-up → #webdev

Fill a PDF form inside a Cloudflare Worker

No Chromium, no Lambda — deterministic AcroForm fill at the edge.

pdfops.dev

Building an AI agent that needs to fill a real PDF form? Give it one tool: POST a field map, get deterministic bytes back — works from a Cloudflare Worker or Lambda. No browser to drive, no binary to bundle. #webdev

PDFops playground

Try fill + merge live, then copy the exact API call.

pdfops.dev

$0.004 per PDF call. No per-seat license, no headless-browser tax, runs from Cloudflare Workers or Node.js. Free tier is 100 requests/month, no signup. How it compares to Anvil → #webdev

PDFops vs Anvil

Edge-native, deterministic, $0.004/call.

pdfops.dev

Merge N PDFs into one with a single HTTP call from any serverless runtime — order preserved, output deterministic. Pairs with fill: build the doc, stamp the fields, ship the bytes. No native deps. #webdev

PDFops — merge & fill PDFs

Deterministic merge + fill from one API.

pdfops.dev

new on daily.dev: filling PDF forms inside a Cloudflare Worker — no Chromium, no Lambda, one fetch call. join the discussion:

Fill a PDF in Node without a headless browser: POST the template plus a {field: value} map, get bytes back. Same call works from a Lambda, a Worker, or a plain box. →

Fill a PDF in Node.js

One fetch, no native deps, no headless browser.

pdfops.dev

Filling a PDF in Python usually means a native lib that won't deploy to serverless. The alternative: one POST with a JSON {field: value} map. Both approaches, side by side →

Fill a PDF in Python

Templates, native libs, and the one-HTTP-call approach compared.

pdfops.dev

You can't run a headless browser in a Cloudflare Worker. So how do you write a PDF there? You POST a template + a field map to something that returns deterministic bytes. That's the entire product. →

PDFops — PDF fill & merge for the edge

One HTTP call. No headless browser. Runs on Workers/Vercel/Deno/Bun.

pdfops.dev

Deterministic PDF output means: same input → byte-identical bytes. So you can hash it, cache it, diff it in CI, and sign it. Most hosted PDF APIs can't promise that. Why it's the whole point →

The case for deterministic PDF filling

Byte-identical output you can hash, cache, and sign.

pdfops.dev

Need to fill a PDF form in code but don't know the field names? Paste the PDF into the Inspector — it lists every AcroForm field, its type, and current value. No signup, nothing to install. →

PDF field Inspector — PDFops

Paste a PDF, see its AcroForm fields. No signup.

pdfops.dev