Rosetta Stone Protocol

Universal cross-domain concept translation. Each lexicon is a domain's language; the engine routes concepts between them via shared universal primitives. Designed to bind authorship on-chain through VibeSwap's PrimitiveRegistry + ContributionDAG + ShapleyDistributor so that citation flows generate royalty attribution by structure, not promise.

Status: v0.1.0 — production-grade installable starting point. Engine is extracted from the working VibeSwap frontend /rosetta route. Server is local-first SQLite (single-node, not multi-tenant cloud). On-chain attribution is spec'd in this repo against deployed VibeSwap contracts — the wiring (PrimitiveRegistry binding + CitationAnchor epoch sealing) is implemented as integration points, not yet end-to-end live.

---

What this is

A monorepo with three packages:

Package	Purpose	Run-mode
@rosetta/engine	Pure ESM translation engine. Lexicons + concept graph + translate / translateToAll / discoverEquivalent / composeLexicon / lexiconShapley.	Embeds in any JS runtime.
@rosetta/server	Node + Express + better-sqlite3 backend. Lexicon registry, CKG snapshots, translation citation logging, epoch sealing.	Local-first SQLite.
@rosetta/web	Vite + React demo SPA. Shows the engine running against the bundled in-memory lexicons.	Localhost dev server.

Plus lexicons/ — the contribution surface for new domain lexicons.

---

Quick start

# 1. Clone
git clone https://github.com/WGlynn/rosetta-stone-protocol.git
cd rosetta-stone-protocol

# 2. Install (npm workspaces)
bash install.sh                        # or just: npm install

# 3. Start the server
npm run dev:server                     # http://localhost:4242

# 4. Start the web demo (in a second terminal)
npm run dev:web                        # http://localhost:3001

Smoke test:

curl http://localhost:4242/health
# -> { "ok": true, "version": "0.1.0", "db": "..." }

---

API

The server exposes three resource families. SQLite schema is created lazily on first boot.

/lexicons — lexicon registry

GET    /lexicons                 list all registered lexicons (metadata)
GET    /lexicons/:id             full lexicon contents
POST   /lexicons                 register a new lexicon (or version-bump an existing one)
PUT    /lexicons/:id/primitive   bind lexicon to its on-chain PrimitiveRegistry tokenId

Lexicon shape:

{
  "id": "oph",
  "domain": "Observer-Patch Holography",
  "author_did": "did:key:z6Mk...",
  "concepts": {
    "observer_patch":  { "universal": "bounded_perspective", "desc": "..." },
    "mismatch":        { "universal": "consistency_break",   "desc": "..." }
  }
}

Content hashing: the server canonicalizes concept JSON (sorted keys) and stores content_hash = sha256(canonical_json). At deploy time this swaps for keccak256 to match the on-chain hash. The keccakHex helper in packages/server/src/routes/lexicons.js is the swap-point.

/ckg — Compiled Knowledge Graph snapshots

GET    /ckg/:snapshot_id         retrieve a saved CKG snapshot
POST   /ckg                      persist a CKG snapshot

/attribution — translation citations + epoch sealing

POST   /attribution/translation     log a translation event (called by every translate())
GET    /attribution/epoch/:id       read a sealed citation epoch
POST   /attribution/epoch/seal      seal accumulated citations into a new merkle epoch

The epoch/seal route aggregates routed_via paths from every logged translation since the last sealed epoch, rolls them up to per-lexicon citation counts, and emits a merkle_root ready to anchor on-chain via CitationAnchor.anchorCitations(merkle_root).

---

On-chain attribution (integration points)

The whole point of the Rosetta Stone Protocol is that citation flow ⇒ authorship attribution by structure, not by promise. Integration with VibeSwap closes that loop. Three contracts do the load-bearing work:

Contract	Role
PrimitiveRegistry	Each lexicon = one Primitive NFT. mint() binds (content_hash, author) to a tokenId. The server's PUT /lexicons/:id/primitive writes the tokenId back.
ContributionDAG	Records the citation graph: lexicon A → lexicon B edges, weighted by translation count from the sealed epochs.
ShapleyDistributor	Computes marginal-contribution rewards across the DAG. Foundational lexicons (cited via many translation paths) earn more by structure — the Cave Theorem.

The integration points are spec'd in this repo against named contracts; the audits workspace tracks the federation + escrow + storage work that this repo plugs into:

• audits/psinet-mindmesh-cycle-1/agent-A-primitive-nft.md — PrimitiveRegistry design
• audits/psinet-mindmesh-cycle-1/agent-B-datatoken-exchange.md — Lexicon datatoken pricing
• audits/psinet-mindmesh-cycle-1/agent-C-mindmesh-federation.md — MindMesh peer publishing
• audits/psinet-mindmesh-cycle-2/agent-D-honest-attestation.md — Author attestation discipline
• audits/psinet-mindmesh-cycle-2/agent-E-privacy-revocation-keyrotation.md — Author key rotation
• audits/psinet-mindmesh-cycle-3/agent-G-escrow-vault.md — Citation royalty escrow

Honest-number: the contract names + their roles are deployed against VibeSwap's existing Primitive / DAG / Shapley stack on testnet; the end-to-end wiring from this server through to a settled royalty payout is not yet executed end-to-end. The server emits the merkle root; the anchor + payout cycle is the next milestone.

---

Upcoming major lexicons

The protocol becomes load-bearing only when authoritative domain experts publish their own lexicons. Two candidate lexicons we'd like to see:

• Observer-Patch Holography (OPH) — Bernhard Mueller's concept graph. Source-of-truth visualization: https://idea-graph-scribe.lovable.app/. OPH would make a strong first cross-domain bridge: its primitives (observer_patch, mismatch, repair, record, holonomy_defect) connect sociology, politics, economics, philosophy, and math under a single mathematical structure. If an OPH lexicon were authored and landed as lexicons/oph.json, the engine's composeLexicon + discoverEquivalent operations would route translations through it wherever a universal concept aligns.

• "Another World Becomes Possible" — Crys's 20-chapter book. A candidate lexicon mapping each chapter's vocabulary into the universal concept space would bring narrative / political / economic primitives into the cross-domain graph.

---

The 4-substrate convergence

Rosetta is one of four substrates Will + JARVIS are extracting from the same structural-property research thread:

1. VibeSwap — omnichain DEX, MEV-free batch auctions, on-chain attribution stack (PrimitiveRegistry + ContributionDAG + ShapleyDistributor)
2. JARVIS-OS — cognition shell + memory + hook gates (github.com/WGlynn/jarvis-os — sibling installable)
3. Rosetta Stone Protocol — this repo — universal cross-domain translation as a public substrate
4. OPH (Observer-Patch Holography) — Bernhard Mueller's mathematical framework, lexicon pending

All four resolve the same underlying property: making structural correctness load-bearing rather than ceremonial. Rosetta is the surface where domain experts compose against each other's vocabularies and get paid by structure for it.

---

Contributing a lexicon

See lexicons/README.md. Short version:

1. Author your domain lexicon as JSON matching the schema
2. PR it into lexicons/, OR POST /lexicons to a running server instance
3. Optional: mint as a Primitive NFT on VibeSwap (PrimitiveRegistry.mint) to bind authorship on-chain and become eligible for Shapley royalty attribution when translations cite your concepts

---

License

MIT. See LICENSE.