SYNAPSE Registry

Canonical IR protocol for AI model interoperability. Write two functions — connect your model to every other model in the ecosystem.

# Without SYNAPSE: N models need N×(N-1)/2 custom connectors connectors = N * (N - 1) / 2 # 10 models = 45 connectors, each breaks on schema change # With SYNAPSE: write ingress() + egress() once connectors = 2 * N # 10 models = 20 adapters, all composable

pip install synapse-adapter-sdk # Write your adapter from synapse_sdk import AdapterBase, CanonicalIR class MyModelAdapter(AdapterBase): MODEL_ID = "my-org/my-model-v1" ADAPTER_VERSION = "1.0.0" def ingress(self, ir: CanonicalIR) -> dict: return {"input": ir.payload.content} def egress(self, output: dict, ir: CanonicalIR, latency_ms: int) -> CanonicalIR: return self.build_response(ir, output["result"], latency_ms) # Validate and check registry in one step synapse-validate --adapter my_module.MyModelAdapter --check-registry