Composable Models for Web3
The web3 ecosystem reinvents the same state machines in every protocol. We're building the parts bin.
We seek useful models across web3 — token standards, governance, DeFi, games — and formalize them in typed JSON-LD. Each model is sealed with a content-addressed CID and anchored in a merkle root. One schema. Verified. Composable.
The Problem
Scattered Standards
ERC-20, ERC-721, ERC-1155, ERC-4626 — each defines its own state machine in English prose. No shared formalism connects them.
Reimplemented From Scratch
Every protocol hand-rolls transfer, approve, mint, burn. Same logic, different bugs, no shared verification.
Unverified Composition
Contracts compose at runtime, but there's no formal model of the composed system. When two protocols interact, you're trusting English docs, not a checkable spec.
Unverifiable Claims
Whitepapers describe behavior in natural language. There's no machine-checkable spec to audit against.
Seek, Formalize, Seal
Seek
Identify useful models across web3 ecosystems: token standards, governance protocols, DeFi primitives, game mechanics.
Formalize
Express each model as typed JSON-LD. Five categorical net types classify every model by the role its tokens play.
Seal
Content-address with a CID, anchor in a registry merkle root. Deterministic, verifiable, immutable.
Networks
Local development and testing. Experiment with models in a sandboxed environment.
Explore →Pre-production validation on Base Sepolia. Verify models before anchoring to mainnet.
Explore →Production registry on Base L2. Sealed models with on-chain merkle root verification.
Explore →Categorical Net Types
Five fundamental types classify every model by the role its tokens play. Each type has distinct structural invariants and composes with others through typed links.
WorkflowNet
Single control-flow token moving through states. Models sequential processes with one source and one sink.
ResourceNet
Countable inventory tokens consumed and produced. Models manufacturing, supply chains, and token economics.
GameNet
Turn-based tokens with player roles. Models adversarial interactions, auctions, and protocol negotiations.
ComputationNet
Continuous quantities evolving via ODE rates. Models chemical kinetics, population dynamics, and signal flow.
ClassificationNet
Signal tokens with threshold activation. Models neural layers, decision trees, and feature routing.
Composition via Links
Typed nets compose into larger systems through four link types. Each link defines how tokens, data, and control flow between subnets.
EventLink
Broadcast signals between nets on transition firing.
DataLink
Share place state as read-only context across nets.
TokenLink
Transfer tokens between places across net boundaries.
GuardLink
Gate transitions on conditions in other nets.
JSON-LD Schema
Models use a semantic web foundation with typed JSON-LD. The @type field declares the categorical net type, enabling tooling to validate structural invariants.
{
"@context": "https://pflow.xyz/schema",
"@type": "WorkflowNet",
"places": {
"ready": { "initial": [1], "capacity": [1] },
"processing": { "initial": [0], "capacity": [1] }
},
"transitions": {
"start": {},
"complete": {}
},
"arcs": [
{ "source": "ready", "target": "start", "weight": [1] },
{ "source": "start", "target": "processing", "weight": [1] },
{ "source": "processing", "target": "complete", "weight": [1] },
{ "source": "complete", "target": "ready", "weight": [1] }
]
}Join the registry. Formalize a model. Verify a seal.