Phases 0 and 1 — the substrate and the operational layer — are complete and verified. Real signature verification, a live Arweave client, and a deployable resolver service are shipped and fail closed. What remains is gated on an external audit and named honestly, phase by phase. This is the roadmap, not a wish list: a box is shipped only if the pipeline exercises it today.
ShippedPartial · defined seamRoadmap · nextVision · sibling scope
Phase 0
The substrate
SHIPPED
The engineering baseline — the part everything else is gated on — is complete and reproducible from one command. This is what an evaluator can run today.
TypeScript SDK with a lean, audited dependency surface: resolveAid, validateAid, verifyChain, canonicalize
JSON Schemas, the RDF ontology, and the knowledge-graph module — projected and proven in Oxigraph (98 quads / 2 named graphs)
The two hash rules and the frozen praxis-canonical-json/1 encoding, with golden byte-equality vectors
CI with an artifact-drift gate proven in both directions · 68 tests · npm provenance on release
Phase 1
The operational layer
SHIPPED
The three boundaries the specification build left as defined seams are now closed and exercised by the pipeline. Real cryptography, real storage, real resolution — each fails closed rather than returning a plausible-but-fake result.
Real signature verification — BLS12-381 (Tezos tz4 min-sig), Ed25519, P-256, secp256k1 via @noble/curves, behind a CryptoProvider boundary; proofs are content-bound (a v1 proof is rejected on v2)
Live Arweave client — multi-gateway fetch with failover and a size ceiling (gateways untrusted), plus a bundler publisher with an HSM/KMS Signer seam
Deployable resolver service — chain → Arweave → re-hash → signature → status, returning an auditable proofChain; holds no keys and cannot mutate state
Signed examples verify end-to-end; runtime surface is two audited dependencies with 0 vulnerabilities in the shipped build
Phase 2
Contract & audit
CURRENT FRONTIER
The soulbound FA2 anchor contract has a full SmartPy test harness exercising every invariant and authority path — but it remains unaudited, and the tooling refuses to originate it to mainnet by policy. This is the one boundary held closed by choice, not capability.
SmartPy scenario suite for every authority path and the two-step controller handoff — present and passing
The five TAP-8 events and fail-closed transfer semantics confirmed against the spec
Independent third-party audit — the gate before any origination
Guarded Ghostnet origination (the deploy path compiles always, refuses mainnet)
Milestone: one real end-to-end identity on Ghostnet, after audit
Phase 3
The hosted resolver network
ROADMAP
The resolver is already deployable — this phase runs it as a hosted, federated network. Convenience over the protocol, never authority within it: independently built indexes federate by merging quads, and each side re-verifies for itself.
Hosted resolver & indexer network with federation by set-union merge
Developer dashboards, analytics, and enterprise integrations
Indexers remain non-authoritative — the anchor and the recomputation are the truth
Phase 4
Standards & registration
ROADMAP
Bind the identifiers and controls to stable, citable homes so the format becomes a standard rather than a single-vendor artifact.
Register the w3id.org/praxis identifiers at stable URLs
Formal alignment with NIST NCCoE, SP 800-63, and OWASP NHI guidance
The conformance evaluator describes compliance for others to certify against
Phase 5
Hardening & mainnet
ROADMAP
Production posture: sustained review, threat-model hardening, and the operational maturity a mainnet identity registry demands before institutions depend on it.
Continuous security review and threat-model expansion
Operational SLAs, monitoring, and incident procedures
Mainnet readiness for institutional identity at scale
Invariant I11 — honesty about boundaries
The status on this page is not editorial. It is drawn from the repository's vision-to-code cross-reference, and CI enforces the distinction: a phase reads SHIPPED only if the pipeline exercises it today. Everything ahead is named, phased, and gated — an investor and an auditor are shown the same map, because a trust system that hides its own edges has not earned trust.