Roadmap to npm

This is the product roadmap: what ships, in what order, on the way to a published npm package you can run with npx proofmark. It covers shipped behavior and planned work only — it makes no claims about features that don't exist yet.

Status legend:

• ✅ Shipped — available today on main.
• 🚧 In progress — partially built, not yet complete.
• 🔭 Planned — designed, not yet built.

Today Proofmark runs from a clone (npm run cli -- ...). The headline goal of this roadmap is the first published release: npx proofmark with no clone.

Now — Alpha (Prisma + Postgres, dry-run)

The current supported path. All of this works today.

Capability	Status
Prisma schema discovery and masking-rule inference	✅ Shipped
Read-only Postgres sampling	✅ Shipped
HMAC-keyed deterministic masking (fails closed without a workspace seed)	✅ Shipped
CLI evidence output with a Proofmark Certificate	✅ Shipped
init generates proofmark.json + a gitignored PROOFMARK_WORKSEED	✅ Shipped
Runnable example Prisma/Postgres project	✅ Shipped

Masking is deterministic, transparent, and stateless — no real-to-fake values are ever stored. See How masking works for the algorithm and the security model, including what a leaked seed does and does not expose.

Next — First npm release

The work that turns "clone and run" into "install and run."

Milestone	Status	Notes
Publish CLI to npm so npx proofmark works without cloning	🔭 Planned	The dry run today is local; this is the gate to public install.
Public demo route that renders real dry-run evidence (not mocked)	🚧 In progress	Example project shipped; the public demo route is still deferred.
Quickstart and CLI reference docs on proofmark.dev/docs	🚧 In progress	This docs site.

What "ready for npm" means

A release is publishable when:

1. npx proofmark init → pull --dry-run works against a real Prisma/Postgres database from a clean machine.
2. A non-demo dry run with no resolvable PROOFMARK_WORKSEED exits non-zero with a clear error (fails closed).
3. npm pack --dry-run ships only the CLI surface — no internal tooling, planning notes, or experimental dashboard code.
4. README and docs describe only shipped behavior, with planned items clearly marked.

Later — After v1

Larger capabilities that extend Proofmark beyond a read-only dry run. These are designed, not built — they are not promised in any current release.

Capability	Status	Notes
Write/seed mode — load masked sample data into a local or target database	🔭 Planned	Today's path is dry-run only; nothing is written to source or target.
Additional schema adapters beyond Prisma	🔭 Planned	The dry-run sampler is Prisma-only today.
Additional source databases beyond Postgres	🔭 Planned
Plugin API for custom maskers	🔭 Planned	Documented as planned in How masking works; not yet implemented.

Out of scope here

Go-to-market plans, launch sequencing, pricing, and funding are tracked privately, not in this repo or on this site. This page stays focused on the product.

Want to influence this?

Open an issue or start a discussion on GitHub. The most useful contribution right now: bring a messy table that makes local testing painful, and a dry run should prove your app still works without keeping the real identifiers.