OpenPubkey is trying to solve a narrower and more foundational problem than most walletless login or artifact signing products: how to bind an OIDC identity to a user- or workload-held public key without introducing a new certificate authority.
The core mechanism is to make the OpenID Provider-signed ID token commit to the user’s public key. For user identities this commitment is embedded through the nonce; for workloads the materials say the same role is played by aud because many workload providers do not return a usable nonce.
The resulting object, a PK Token, functions like a certificate chain compressed into OIDC-native parts: OP-authenticated identity claims, a key commitment, and a user proof-of-possession signature.
The FAQ is especially valuable because it makes clear that OpenPubkey is not a replacement for Sigstore or similar systems. It is the lower layer that binds identities to keys; a larger signing, transparency, or policy system can sit above it.
Public use brings a subtle but important problem: the raw OP-signed ID token can become a replayable bearer artifact. The FAQ and README say OpenPubkey now integrates GQ-signature-based proofs so the OP signature can be stripped and replaced with a proof, preserving the authentication assurance while reducing replay risk.
Long-term verification is another explicit control surface. Because OIDC providers rotate JWKs, any archival verification path depends on some history of past issuer keys. The project materials discuss several approaches, including archival verifiers, TUF-distributed key histories, and transparency-log-like ledgers.
The public overview page usefully frames OpenPubkey as reusing the OIDC provider in the role a CA would normally occupy, while leaving the user or workload in control of the signing key.
Specific source signals worth retaining
The README is the clearest source for the protocol decomposition: ID token → committed key → PK Token → verifier extraction and signature verification.
The FAQ is the clearest source for the boundaries of the project: complementary to Sigstore, focused on identity-to-key binding rather than complete artifact-signing workflow.
The FAQ also captures the most important operational caveats: public-claim leakage inside PK Tokens, replay risk for public ID tokens, and long-term verification problems created by OP key rotation.
The Web Encrypt overview is useful because it ties the protocol to concrete deployment surfaces like Docker image signing and OpenPubkey SSH while still describing the underlying trust shift away from a separate CA.
The IACR paper is the best concise statement of the project’s central thesis: upgrading OIDC from bearer authentication to proof-of-possession without requiring changes from existing OpenID Providers.
Why this entry belongs in the corpus
It gives the corpus a lower-layer comparison point beneath social-login wallets, artifact-signing systems, DID products, and account-abstraction authentication flows.
It helps separate identity-provider trust, key binding, proof-of-possession, replay defense, and historical issuer-key custody instead of flattening them into one generic passwordless auth or walletless onboarding story.
The durable insight is that many systems described as sign in with X or sign with your existing account are really bundles of several distinct layers, and OpenPubkey makes those layers unusually legible.