Everything in Pulp Engine

Built for developers and engineering teams. Grouped by who tends to care. Every item ships under the same licence — no Pro tier, no feature gates.

For developers

Ship document features without inventing infrastructure.

REST + OpenAPI 3.0

Full spec at /docs/json. Swagger UI at /docs. Generate your own client in any language.

Typed SDKs

TypeScript, Python (Pydantic v2), Go, .NET — version-locked to the server release.

CLI tooling

@pulp-engine/cli: push/pull for content sync, render/dev for local iteration, validate/test/diff/merge, promote, codegen, and backup.

<code>dryRun: true</code>

Validate templates and exercise expressions without producing output. ~80× faster than a full render.

Plugin system

Add custom node types, renderers, storage backends, auth providers, render hooks, or new API routes.

Async batch rendering

POST /render/batch/async (PDF and DOCX) returns a job id, delivers results via webhook callback, and survives restarts with a durable DLQ.

Scheduling & delivery

Cron-scheduled renders with pluggable delivery: email (SMTP), S3/object storage, or webhook. Retry policy, DLQ inspection, and per-schedule audit trail.

PDF post-processing

POST /pdf/merge, /pdf/watermark, and /pdf/insert for combining, stamping, or splicing PDFs without re-rendering from a template.

Observability built in

Prometheus metrics, structured logs, OpenTelemetry, X-Request-ID propagation, render accounting headers.

Bill your customers per render

Metered usage out of the box: GET /usage + /usage.csv admin endpoints (per-tenant counts, durations, bytes, pages), render:completed plugin event for real-time billing pipelines, and X-Render-* response headers on every render. Pipes cleanly to Stripe Metering, Orb, Metronome, or your own invoice job. Rollup endpoints require a database-backed storage mode (Postgres or SQL Server); headers and plugin events work everywhere.

For engineering teams

Authoring, versioning, and delivery surfaces built for developer authors — not business users.

Visual template editor

Rich text, tables, 16 chart types, images, conditionals, repeaters, pivot tables, barcode/QR, signature fields, and table-of-contents. Lives at /editor/.

Six output formats

PDF (Chromium), DOCX, PPTX, XLSX, CSV, HTML — from one template definition.

AI template scaffolding

Prompt → draft TemplateDefinition, schema-valid by construction. A shortcut for devs starting a new template, not end-user magic.

Version history + diff

Every save is a version. Restore, promote labels (e.g. stable) without redeploying, and POST /templates/:key/diff for structural diffs between any two versions — also callable from CI.

Embeddable editor + form

<pulp-engine-editor> drops the full editor into a customer app via iframe. <pulp-engine-form> exposes just the data-entry surface for end users.

Starter templates

28 built-in starting points in the new-template dialog, from Blank and invoices to pivot examples, component showcases, page chrome, and stock reports.

For ops & security

Run it like any other production service.

Three storage backends

Postgres (default), SQL Server, file-backed. Swap with one env var.

S3 asset binaries

Object-storage-backed asset binaries with health probes. Multi-instance without shared volumes.

Multi-tenant mode

MULTI_TENANT_ENABLED isolates templates, assets, credentials, audit events, and schedules per tenant.

OIDC / SSO

PKCE auth-code flow with auto-provisioning and silent refresh. Okta, Azure AD, Keycloak, Auth0.

Scoped API credentials

Separate keys for admin, render, preview, and editor. Verify-only previous-key rotation slot.

Runtime user management

/admin/users CRUD plus /admin/users/reload — provision editors, rotate attribution, and re-read the EDITOR_USERS_JSON registry without restarting the API.

Evaluation & licensed builds

Same binary, one env var: unlicensed deploys stamp a small watermark on output (fine for development, staging, and internal PoCs). A paid PULP_LICENCE_KEY removes it — no separate "enterprise" build.

Render isolation modes

in-process · child-process (default) · container · socket (privilege-separated).

Hardened production posture

Docs disabled by default, CSP, HTTPS enforcement, magic-byte upload validation, DNS-rebinding SSRF guard.

Audit trails

Per-actor audit events, named-user attribution, retention-aware purge with erasure-by-actor.

HA + backup

Reference compose file, nginx upstream config, backup CLI with verify, restore runbook.

Honest heads-up Known limits · being actively addressed

Where Pulp Engine doesn't pretend yet

A new commercial product deserves honest limits. These are the questions larger-team buyers ask before rolling out production deployments; none are blockers for most teams today, and Early Adopters get direct input as we close them.

Per-template RBAC

Not yet

Access control today is by scope (admin · editor · render · preview) and, in multi-tenant mode, by tenant. There is no per-template permission layer — any editor in a tenant can edit any template in that tenant. Fine for most teams; worth knowing if your compliance model requires "template owner" / "read-only reviewer" roles. On the roadmap.

Load benchmarks

Reproducible harness (numbers pending first run)

A six-cell render benchmark — warm PDF at three page-count sizes (1 / 10 / 100) plus DOCX / PPTX / XLSX at the 10-page template — with a reproducible compose stack, load script, and three template packs. Render-mode comparison (child-process · container · socket) and HA scaling (1 / 2 / 4 instances) are covered by re-running the harness against the matching stack configuration. The harness ships in the repo so you can run it on your own hardware; the first published numbers on the reference rig are still pending. See the benchmark pack.

Render process isolation

Opt-in, not default

RENDER_MODE=child-process (default) shares a Chromium pool across tenants. For stricter tenant-level rendering isolation, RENDER_MODE=container or socket give you per-render sandboxing. Default is fine for most; regulated deployments should pick a stronger mode.

Ready to try it?

Free evaluation runs in under a minute. No database, no signup, no credit card.

Start free evaluation Pricing from USD $399/year