Bronze Scribe

Bronze Scribe is the final storage agent in the pipeline. It takes clean, classified, quality-verified data and writes it to the Bronze layer of your data lakehouse using ACID transactions. It supports Apache Paimon, Apache Hudi, and Apache Iceberg table formats, and can write to any object storage backend (S3, GCS, Azure Blob, MinIO).

Capabilities

Apache Paimon ACID writes
Apache Hudi ACID writes
Apache Iceberg ACID writes
S3, GCS, Azure Blob, MinIO support
Partition management
Compaction scheduling
Write metrics reporting

Configuration

bronze-scribe.yaml

agent: bronze-scribe
version: "1.0"
table_format: paimon  # paimon | hudi | iceberg
storage:
  backend: s3
  bucket: "${BRONZE_BUCKET}"
  prefix: bronze/
partitioning:
  strategy: date  # date | source | custom
  date_column: ingestion_date

Example

bronze-scribe-example.ts

const result = await mcp.call('bronze_scribe_write', {
  payload: qualityVerifiedPayload,
  stream_uuid: streamUUID,
  table_name: 'salesforce_contacts_bronze',
  table_format: 'paimon',
});
console.log(result.rows_written);  // 49823
console.log(result.table_path);    // "s3://bronze/salesforce_contacts/"

Pricing

$0.000010 USDC/row

View full agent page →