Skip to content

Roadmap

The plan is public and small. Each milestone delivers something a person can use, and each work item is an issue with a “done when”. Dates are not promised; the order is.

Status on 2026-11-01: design phase. The scope is decided in ADR 0001, the concepts are written, and the site is being set up. No milestone is started.

The first version supports three key stores: software, YubiKey and AWS KMS. It is built in four milestones.

Milestone Delivers Work items
M1: Hand-off and software roots A system that holds intermediates can get a certified intermediate or a bound subkey from a software root, using standard formats. Nothing else is needed to start. The hand-off contract specification. Software-root mode: create a root, sign a CSR, bind a subkey. Ceremony records, first version. The Go module and CLI skeleton.
M2: Hardware and cloud roots The same ceremonies with the root on a YubiKey or in AWS KMS, with proof of where intermediate keys live. The YubiKey back end (PIV and OpenPGP applets, attestation chain to the maker’s current root, the policy and serial extensions). The AWS KMS back end, with an MFA-gated role as the “touch”.
M3: Transitions and trust products A root can be replaced without breaking anyone, and outsiders get standard artifacts. Vouch, overlap, supersede and revoke, including revocation certificates made in advance and designated revokers. The trust bundle and the signed root list. The keyring builder for .deb, .rpm and plain files.
M4: Wizards, personal GPG and teaching Adopting ritueel is guided, and the teaching material is complete. The CLI wizard. The localhost web wizard. Personal GPG flows. The guides, one per transition. The drills.
  • M1 comes first because the first consumer needs a software root before anything else exists. It is also the cheapest store to teach with.
  • The hand-off contract is the first item in M1: it is specified once and tested from both sides, so that the two ends can be built independently.
  • M2 follows because the real roots live on hardware or in a cloud service. The attestation checks are the part with the most surprises, so they get a milestone of their own.
  • M3 needs M2 to be meaningful: a transition between stores needs two stores.
  • M4 comes last because the wizards drive commands that must exist first. The site and the concepts are not waiting for it; they are written alongside.
  • Company-wide GPG management, with its own decision record.
  • More key stores, each with a record: other clouds, HSMs, TPMs and other cards.

None of these is planned in detail. Each starts as a decision proposal.

The milestones are GitHub milestones on the repository. The issues carry the labels area:* and type:*. A change to this plan is a pull request to this page.