Implementation:Kubeflow Pipelines MLMD Service Protobuf

Knowledge Sources	Kubeflow_Pipelines ML Metadata
Domains	Metadata_Management, API_Schema
Last Updated	2026-02-13 14:00 GMT

Overview

Auto-generated Go protobuf bindings defining all request/response message types for the ML Metadata Store gRPC service API.

Description

Generated by `protoc-gen-go` (v1.26.0) from `ml_metadata/proto/metadata_store_service.proto`, this is the largest generated file in the repository (10,207 lines). It defines Go structs for every MLMD API operation's request and response messages, organized into CRUD patterns: Put operations (PutArtifacts, PutExecutions, PutEvents, PutContexts, PutTypes, PutLineageSubgraph) and Get operations (GetArtifactsByType, GetExecutionsByID, GetContextsByArtifact, GetLineageGraph, GetEventsByExecutionIDs, etc.).

Usage

Imported by the KFP Go backend alongside the gRPC stubs to make type-safe MLMD API calls for recording and querying pipeline metadata.

Code Reference

Source Location

• Repository: Kubeflow_Pipelines
• File: third_party/ml-metadata/go/ml_metadata/metadata_store_service.pb.go
• Lines: 1-10207

Signature

package ml_metadata

type PutArtifactsRequest struct {
    Artifacts []*Artifact
    Options   *PutArtifactsRequest_Options
}

type PutArtifactsResponse struct {
    ArtifactIds []int64
}

type GetArtifactsByTypeRequest struct {
    TypeName *string
    Options  *ListOperationOptions
}

type GetArtifactsByTypeResponse struct {
    Artifacts     []*Artifact
    NextPageToken *string
}
// ... 80+ request/response message types

Import

import ml_metadata "github.com/kubeflow/pipelines/third_party/ml-metadata/go/ml_metadata"

I/O Contract

Inputs

Name	Type	Required	Description
Proto descriptor	[]byte	Yes	Raw protobuf descriptor from metadata_store_service.proto

Outputs

Name	Type	Description
Put request types	Go structs	PutArtifactsRequest, PutExecutionsRequest, PutContextsRequest, etc.
Put response types	Go structs	PutArtifactsResponse (with generated IDs), etc.
Get request types	Go structs	GetArtifactsByTypeRequest, GetExecutionsByIDRequest, etc.
Get response types	Go structs	GetArtifactsByTypeResponse (with entities + pagination), etc.

Usage Examples

Putting Artifacts via Message Types

import ml_metadata "github.com/kubeflow/pipelines/third_party/ml-metadata/go/ml_metadata"

req := &ml_metadata.PutArtifactsRequest{
    Artifacts: []*ml_metadata.Artifact{
        {Uri: proto.String("gs://bucket/data"), TypeId: proto.Int64(1)},
    },
}
// Pass req to gRPC client stub

Related Pages

• Environment:Kubeflow_Pipelines_KFP_Backend_Deployment