protobuf.ai — Protocol Buffers for the AI Era

Stop AI from guessing your schema semantics

When AI agents integrate with your API, they see field names and types. They don't know that amount is in cents, that status follows a state machine, or that refunds use negative values.

context.json gives every AI tool a single, always-current source of truth for what your schema actually means—so integrations work correctly without reverse-engineering your documentation.

AI-generated integrations work correctly by default—no trial and error.

Update semantic context once—every AI agent picks it up instantly.

Ontology mapping validates that fields match their intended meaning.

Works alongside your existing Buf, protoc, and gRPC tooling.

One file. Complete semantic context.

Everything AI agents need to understand your API—organized, versioned, and always up to date.

AI Discovery

When developers ask "what should I use for payments?", AI agents find your API through semantic search—not keyword matching.

Correct Implementation

AI understands that amount is in cents, currency is ISO 4217, and refunds use negative values. First integration works.

Ontology Validation

Map fields to schema.org or custom ontologies. Catch semantic mismatches before they become production bugs.

Cross-Schema Compatibility

Two APIs with different field names but the same meaning? context.json knows they're compatible.

Strategy-as-Code

Your schema semantics, version-controlled and API-accessible.

// payment.context.json
{
  "@context": "https://protobuf.ai/ontology/v1",
  "schema": "payment.proto",
  "domain": "payments",
  "messages": {
    "Payment": {
      "semanticType": "schema:PayAction",
      "fields": {
        "amount": {
          "maps_to": "schema:price",
          "unit": "cents"
        }
      }
    }
  }
}

Edge Cases Captured

Domain knowledge that lives with the schema, not in tribal knowledge.

// payment.context.json
{
  "edge_cases": [
    "Refunds use negative amount",
    "Multi-currency requires conversion at capture time",
    "Status transitions: PENDING → COMPLETED | FAILED"
  ],
  "validation": {
    "amount": "Must be non-zero",
    "currency": "ISO 4217 format required"
  }
}

How it works

From schema to AI-native API in four steps.

1. Generate

AI analyzes your .proto and creates context.json automatically

2. Refine

Add domain knowledge, edge cases, and relationships

3. Publish

Push to registry.protobuf.ai alongside your schema

4. Discover

AI agents find and correctly understand your API

Make your APIs AI-native

Stop letting AI guess your schema semantics. Give it the truth.

context.json turns your Protocol Buffers into a living system that every AI tool can read—so integrations stay accurate, consistent, and correct.

View on GitHub Explore the Registry