Status: public comparison guide for @bilig/headless.
This page is for developers already evaluating HyperFormula or another headless spreadsheet engine. It is intentionally not a takedown. HyperFormula is the established TypeScript engine in this category; its official README positions it as a headless spreadsheet for business web apps with formula evaluation, CRUD operations, undo/redo, clipboard support, sorting, Node.js support, and a GPLv3 or commercial license.
@bilig/headless is worth evaluating when the workload is closer to a
service-side WorkPaper runtime: formula-backed business logic, structural edits,
agent writeback, persistence, restore, and auditable benchmark evidence from the
same repository.
Use HyperFormula when you want a mature, UI-independent formula calculation engine with broad spreadsheet-function coverage and commercial support from the Handsontable team.
Use @bilig/headless when you need a MIT-licensed Node package that treats the
workbook as a service object: build it, mutate it, read formulas and values,
persist it, restore it, and verify agent-style edits without opening a browser
grid.
| Question | HyperFormula | @bilig/headless |
|---|---|---|
| Primary shape | Headless spreadsheet formula engine | Headless WorkPaper workbook facade |
| Runtime target | Browser or Node.js | Node services, tests, agents, and local runtime automation |
| License posture | GPLv3 or commercial license | MIT |
| API orientation | Spreadsheet-engine instance with formula evaluation and workbook operations | WorkPaper object with formula evaluation, structural edits, persistence helpers, history, and readback |
| Agent workflow fit | Possible, but the project is not specifically packaged around agent writeback proofs | First-class evaluation path includes an agent writeback demo with persistence and restored readback |
| Benchmark claim in this repo | External comparison target | Checked-in WorkPaper-vs-HyperFormula artifact records 46/46 mean wins on scorecard-eligible comparable workloads |
| Caveat | Strong default engine, with its own licensing and integration model | Not a finished Excel clone and not full Excel formula parity |
The current checked-in artifact is:
packages/benchmarks/baselines/workpaper-vs-hyperformula.json
The short benchmark explainer is:
docs/what-workpaper-benchmark-proves.md
The current public claim is narrow:
46/46 mean-latency wins on scorecard-eligible comparable workloads38/38 public-lane mean wins8/8 holdout-lane mean winsThe verification command is:
pnpm workpaper:bench:competitive:check
This does not prove that bilig is faster at every possible spreadsheet task. It does not prove full Excel compatibility. It proves the checked-in WorkPaper runtime claim for the current comparable headless workload scorecard.
Use the published package when you want a quick local evaluation:
mkdir bilig-headless-eval
cd bilig-headless-eval
npm init -y
npm pkg set type=module
npm install @bilig/headless
Use the maintained example when you want an end-to-end proof:
git clone https://github.com/proompteng/bilig.git
cd bilig/examples/headless-workpaper
npm install
npm start
npm run agent:verify
The agent verifier records the assumption cells changed, checks dependent formula readback, persists the workbook, restores it, and verifies the restored values.
For those cases, start with HyperFormula or a full spreadsheet product, then use bilig’s compatibility notes to decide whether a narrower WorkPaper runtime fits a later slice.
packages/headless/README.mddocs/public-adoption-kit.mddocs/what-workpaper-benchmark-proves.mddocs/headless-workpaper-benchmark-evidence.mddocs/where-bilig-is-not-excel-compatible-yet.mddocs/starter-issues.md