Connectors: what works today, what is next, and what is on the roadmap

Read-only connectors that run inside your agent.

Every connector runs inside the customer-resident agent and reads sources read-only. This list is generated from the capability registry the product uses, and each status is derived from recorded conformance evidence: a connector is only called available when the agent ships it and it passed every check.

Find a connector

Showing all 34 connectors.

Browse by category: Databases · Data warehouses · Lakehouses · Object storage · Files · NoSQL · SaaS applications · Mainframe · Streaming · APIs.

Available now (27)

Shipped in the standard agent and backed by recorded conformance evidence: read-only proof, secret redaction, discovery, bounded streaming, egress canaries and the full discover, mask, subset, certify and provision workflow.

Available now NoSQL

Amazon DynamoDB

Shipped in the standard agent and backed by recorded conformance evidence.

Amazon DynamoDB connector (boto3): table discovery, partition and sort keys, attribute shapes inferred from a bounded scan inside your network, and eventually consistent, paginated scans under a read-capacity budget with nested attributes carried as JSON text.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: IAM access key (secret reference); default AWS credential chain (instance profile, workload identity)
Tested environments: Amazon DynamoDB service
Last verified: 2026-10-12 · connector 0.1.0

Known limitations (5)

Requires agent 0.4.0 or later (its driver, boto3, ships in the standard agent image).
Run against one real AWS account (us-east-1, on-demand tables, IAM-user reader proven read-only by policy simulation) and DynamoDB Local; provisioned-capacity tables, global tables, resource-policy grants and assumed-role principals are untested.
Attributes are inferred from a bounded scan sample; a later item whose attribute type differs stops the read with FIELD_TYPE_DRIFT.
Exact counts scan the table under the read budget; discovery reports DynamoDB's ItemCount, which is refreshed about every six hours.
DynamoDB declares no relationships: they come from industry packs or explicit mappings.

Details and docs for Amazon DynamoDB · Conformance evidence for Amazon DynamoDB

Available now Data warehouses

Amazon Redshift

Shipped in the standard agent and backed by recorded conformance evidence.

Dedicated Redshift connector on the PostgreSQL protocol with Redshift's own system views: svv_tables/svv_columns discovery, declared keys, svv_table_info row estimates, bounded aggregate profiles and server-side-cursor reads into the local engine, over verified TLS.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: database user and password (secret reference)
Tested environments: Amazon Redshift service
Last verified: 2026-10-12 · connector 0.1.0

Known limitations (5)

Requires agent 0.4.1 or later: earlier agents cannot prove a Redshift credential read-only (its driver, psycopg, ships in the standard agent image).
Database user and password only; IAM temporary credentials and identity-provider federation are not supported yet.
Primary and foreign keys are informational in Redshift (never enforced); relationships are only as good as the declarations.
External (Spectrum) and datashare objects are not read.
Run against a real Redshift Serverless workgroup (8 RPU); provisioned clusters are not yet tested. No local engine exists and PostgreSQL runs are not accepted as Redshift evidence.

Details and docs for Amazon Redshift · Conformance evidence for Amazon Redshift

Available now Object storage

Amazon S3 / S3-compatible storage

Shipped in the standard agent and backed by recorded conformance evidence.

Built-in file-set connector for Parquet or CSV objects under a bucket prefix in Amazon S3 or an S3-compatible store (endpoint override).

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: access key (secret reference); instance/workload identity
Tested environments: SeaweedFS S3 gateway (S3-compatible) 4.48; Amazon S3 service (amazon-s3-service)
Last verified: 2026-10-08 · connector 0.1.0

Known limitations (3)

Verified against Amazon S3 (a real bucket, least-privilege reader) and an S3-compatible server (SeaweedFS S3 gateway); other S3-compatible stores are not individually tested.
Only access-key authentication is verified; instance and workload identity are not yet part of the conformance run.
Read-only access cannot be proven for object stores: it is attested by the operator, and the source is refused without that attestation.

Details and docs for Amazon S3 / S3-compatible storage · Conformance evidence for Amazon S3 / S3-compatible storage

Available now Streaming

Apache Kafka

Shipped in the standard agent and backed by recorded conformance evidence.

Kafka connector: declared topics read as a bounded snapshot window (log start or a start time up to the last stable offset when the job starts), with JSON fields, text or binary values, record metadata columns and offset-range row counts. It never joins a consumer group and never commits offsets.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: SASL/SCRAM-SHA-512 or SCRAM-SHA-256 (secret reference); SASL/PLAIN over TLS
Tested environments: Apache Kafka 4.1.0
Last verified: 2026-10-08 · connector 0.1.0

Known limitations (7)

Requires agent 0.4.0 or later; earlier agent images do not include the confluent-kafka driver.
Reads a bounded snapshot window per job; it never follows a live stream.
Row counts are offset ranges: exact for ordinary records, an upper bound when compaction or transaction markers occupy offsets.
Topics declare no keys: subsetting and relationships come from industry-pack templates or declared relationships and are closed only within the snapshot window.
JSON field shape is inferred from a bounded sample or declared; Avro and Protobuf with a Schema Registry are not decoded (read such topics as binary).
SASL authentication only; mutual-TLS client certificates are not supported yet.
Masked output goes to the configured provisioning target; nothing is produced back to Kafka.

Details and docs for Apache Kafka · Conformance evidence for Apache Kafka

Available now Object storage

Azure Blob Storage / Data Lake Storage

Shipped in the standard agent and backed by recorded conformance evidence.

Built-in file-set connector for Parquet or CSV files in an Azure Blob container or ADLS Gen2 filesystem, with account-key, SAS or service-principal authentication.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: account key; SAS token; service principal
Tested environments: Azure Storage (Blob / Data Lake Storage Gen2) service
Last verified: 2026-10-06 · connector 0.1.0

Known limitations (3)

Verified against a real Azure Storage account with hierarchical namespace (Data Lake Storage Gen2) using a read + list SAS; account-key and service-principal authentication were verified against Azurite, Microsoft's local emulator, only.
SAS tokens as copied from the Azure portal or CLI (without a leading '?') need agent 0.2.0 or later; agent 0.1.0 accepts them only with the leading '?'.
Read-only access cannot be proven for object stores: it is attested by the operator, and the source is refused without that attestation.

Details and docs for Azure Blob Storage / Data Lake Storage · Conformance evidence for Azure Blob Storage / Data Lake Storage

Available now NoSQL

Azure Cosmos DB

Shipped in the standard agent and backed by recorded conformance evidence.

Azure Cosmos DB for NoSQL connector (azure-cosmos): container discovery, field shapes inferred from a bounded sample inside your network, partition-key-aware keys, RU-budgeted paged queries and nested items carried as JSON text.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: read-only account key (secret reference)
Tested environments: Azure Cosmos DB for NoSQL service
Last verified: 2026-10-12 · connector 0.1.0

Known limitations (6)

Run against one real Azure Cosmos DB for NoSQL account (provisioned throughput, single region) and the Linux emulator; serverless, autoscale and multi-region accounts are untested.
Requires agent 0.4.0 or later; earlier agent images do not include the azure-cosmos driver.
API for NoSQL only; account keys only (Microsoft Entra ID data-plane roles are not supported).
Read-only access is attested, not verified.
Fields are inferred from a bounded sample; a later item whose field type differs stops the read with FIELD_TYPE_DRIFT.
Cosmos DB declares no relationships: they come from industry packs or explicit mappings.

Details and docs for Azure Cosmos DB · Conformance evidence for Azure Cosmos DB

Available now Data warehouses

Azure Synapse Analytics

Shipped in the standard agent and backed by recorded conformance evidence.

Dedicated and serverless SQL pool connector over TDS (the SQL Server connector's driver and TLS enforcement): catalog discovery, declared primary keys, distribution metadata, row estimates from distribution statistics, bounded profiles and reads, labelled for cost attribution.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: SQL authentication (secret reference)
Tested environments: Azure Synapse Analytics (dedicated SQL pool) service; Azure Synapse Analytics (serverless SQL pool) service
Last verified: 2026-10-12 · connector 0.1.0

Known limitations (9)

Requires agent 0.4.1 or later: agent 0.4.0 cannot prove a dedicated-pool credential read-only and does not bind industry-pack relationship mappings to tables in your own schema (its driver, python-tds, ships in the standard agent image).
Azure Synapse has no foreign keys, so DataNivra cannot discover relationships from Synapse: they come from an industry pack's relationship mappings or relationships you declare. Without declared relationships, subsets do not follow relationships and the orphan check cannot cover them (the certificate records RELATIONSHIPS_NOT_DECLARED).
Pack relationship mappings bind to a table in your own schema (for example dbo.claims) only when exactly one table of that name exists in the dataset's sources and it has every column the mapping names; otherwise nothing binds and the console warns before the build. Relationships you declare yourself are approved identity-graph links between Synapse and another source; two Synapse tables are linked through a pack's relationship mappings.
Primary keys are declared NOT ENFORCED: DataNivra reads them but cannot rely on Synapse to keep them unique.
SQL authentication only; Microsoft Entra ID (service principal, managed identity) is not supported yet.
Serverless pools keep no row statistics: counts and profiles read the underlying files and are billed per data processed.
Run against a real serverless SQL pool (external tables over Parquet) and a real DW100c dedicated SQL pool. No emulator exists and SQL Server runs are not accepted as Synapse evidence.
Serverless external tables declare no primary keys, so a dataset build rooted on one stops with ROOT_ENTITY_WITHOUT_PRIMARY_KEY (seen on the real run); the full workflow was proven on a dedicated pool.
Serverless external tables also need REFERENCES on the database scoped credential they read through (see the least-privilege guide).

Details and docs for Azure Synapse Analytics · Conformance evidence for Azure Synapse Analytics

Available now Databases

CockroachDB

Shipped in the standard agent and backed by recorded conformance evidence.

CockroachDB through the PostgreSQL connector, with its own compatibility record: catalog discovery, declared primary and foreign keys, row estimates from CockroachDB's table statistics, bounded aggregate profiles and server-side-cursor reads over verified TLS. PostgreSQL wire compatibility is never assumed: CockroachDB is detected and tested separately.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: SQL user and password over TLS (secret reference)
Tested environments: CockroachDB 26.3.2
Last verified: 2026-10-13 · connector 0.1.0

Known limitations (6)

Administrator check before connecting: the reader's own login cannot see its legacy role options (CREATELOGIN, CONTROLJOB, CANCELQUERY, CONTROLCHANGEFEED, MODIFYCLUSTERSETTING), so the read-only proof does not cover them. An administrator runs SELECT username, options, member_of FROM [SHOW ROLES] WHERE username = 'tdm_reader'; once and expects empty options ({}) for the reader and for any role it is a member of. Everything else is verified automatically.
Requires agent 0.4.2 or later: earlier agents cannot prove a CockroachDB user read-only and refuse the source (the driver, psycopg, ships in the standard agent image).
Password authentication only (no client-certificate, GSSAPI or JWT authentication).
Row estimates need table statistics (CockroachDB collects them automatically); without them the connector counts rows exactly.
Byte-size estimates are not reported.
Tested against a self-hosted CockroachDB 26.3 single node; CockroachDB Cloud (Basic, Standard, Advanced) and multi-region clusters are not yet tested.

Details and docs for CockroachDB · Conformance evidence for CockroachDB

Available now Lakehouses

Databricks

Shipped in the standard agent and backed by recorded conformance evidence.

Databricks connector: Unity Catalog catalog.schema.table discovery of Delta tables and views through a Databricks SQL warehouse, informational keys and bounded Arrow reads.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: OAuth M2M service principal (secret reference); personal access token (secret reference)
Tested environments: Databricks SQL warehouse (Unity Catalog) service
Last verified: 2026-10-07 · connector 0.1.0

Known limitations (5)

Verified against a real workspace (Unity Catalog, serverless SQL warehouse) with OAuth machine-to-machine; personal access tokens are accepted but not part of the recorded run.
Requires agent 0.3.0 or later; earlier agent images do not include its driver.
Unity Catalog does not enforce foreign keys; informational constraints are read where declared.
Workspace and metastore administrator rights are invisible to the read-only proof: use a dedicated non-admin service principal.
Tags are read locally only; external locations, volumes, Delta Sharing and Jobs are not supported.

Details and docs for Databricks · Conformance evidence for Databricks

Available now Databases

Generic SQL (SQLAlchemy)

Shipped in the standard agent and backed by recorded conformance evidence.

Built-in connector for a SQLAlchemy URL: metadata, counts, profiles and streaming reads, with a statement guard that blocks anything other than SELECT/WITH/SHOW/EXPLAIN.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: SQLAlchemy URL (secret reference)
Tested environments: PostgreSQL (SQLAlchemy dialect) 16.2
Last verified: 2026-10-08 · connector 0.1.0

Details and docs for Generic SQL (SQLAlchemy) · Conformance evidence for Generic SQL (SQLAlchemy)

Available now Data warehouses

Google BigQuery

Shipped in the standard agent and backed by recorded conformance evidence.

BigQuery connector: dataset and table discovery through the tables API, partition and clustering metadata, row counts from table metadata, and reads through tabledata.list, which is not billed as query scanning.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: service-account key (secret reference); workload identity / application default
Tested environments: Google BigQuery service
Last verified: 2026-10-07 · connector 0.1.0

Known limitations (4)

Verified against a real Google Cloud project with service-account keys. A key is limited to the read-only scope, which cannot run query jobs: tables are read and counted without jobs, but views need application-default credentials (for example the agent's attached service account).
Requires agent 0.3.0 or later; earlier agent images do not include its driver.
BigQuery primary and foreign keys are unenforced; relationships may need explicit mappings.
Profiles are computed by the agent over a bounded sample of listed rows.

Details and docs for Google BigQuery · Conformance evidence for Google BigQuery

Available now APIs

HTTP APIs

Shipped in the standard agent and backed by recorded conformance evidence.

Generic HTTP/JSON API connector: you declare each resource in a schema contract (path, where the records are, id fields, field types or a JSON Schema, relationships, pagination) and the agent reads it with GET-only requests, rate limits and bounded reads inside your network.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: bearer token (secret reference); basic (secret reference); API key header (secret reference); OAuth 2.0 client credentials (secret reference)
Tested environments: PostgREST 13.0.4
Last verified: 2026-10-08 · connector 0.1.0

Known limitations (6)

Read-only access is attested by the operator, not verified by the connector.
Keys and relationships come only from the schema contract. Resources without declared fields are sampled for field names and types, and only identifier-shaped names are accepted.
JSON responses only. Nested objects and arrays are carried as JSON text; fields below the top level of a record are not flattened.
Record counts page through the whole resource, bounded by max_pages and max_records.
Verified against PostgREST (a REST API server over PostgreSQL). Other APIs depend on an accurate schema contract.
Requires agent 0.4.0 or later; earlier agent images do not include this connector.

Details and docs for HTTP APIs · Conformance evidence for HTTP APIs

Available now Databases

IBM Db2

Shipped in the standard agent and backed by recorded conformance evidence.

Db2 for Linux, UNIX and Windows connector using IBM's driver installed by the customer: SYSCAT catalog discovery, primary and foreign keys, statistics-based row estimates, bounded profiles and streaming reads into the local engine, over TLS by default.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Driver: Customer-installed: ibm_db under the IBM License Information L-WNKS-F8PR3U for IBM Data Server Driver for ODBC and CLI 12.1.4 (ibm_db itself: Apache-2.0)
Authentication: user and password (secret reference)
Tested environments: IBM Db2 Community Edition (container) DB2 v11.5.9.0
Last verified: 2026-10-13 · connector 0.1.0

Known limitations (5)

Customer-installed driver: the standard agent image does not include ibm_db. Build the DataNivra agent extension image (agent/runtime/extensions/ibm-db2), which installs ibm_db and IBM's CLI driver under IBM's licence terms that you accept; without it the agent reports CONNECTOR_NOT_AVAILABLE.
Linux/amd64 agents only: IBM publishes no Linux arm64 CLI driver, so an arm64 agent reports CONNECTOR_PLATFORM_UNSUPPORTED.
User and password authentication only (no Kerberos, LDAP plug-ins or trusted contexts).
Db2 for LUW only; Db2 for z/OS and Db2 for i are not supported. Connecting to them through IBM's driver would also need your own Db2 Connect licence, which DataNivra never ships.
XML columns are read as serialized text; structured and spatial user-defined types are refused with a reason code.

Details and docs for IBM Db2 · Conformance evidence for IBM Db2

Available now Databases

IBM Informix

Shipped in the standard agent and backed by recorded conformance evidence.

Informix connector over DRDA using IBM's driver installed by the customer: system catalog discovery, primary and foreign keys, statistics-based row estimates, bounded profiles and streaming reads into the local engine, over verified TLS by default.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Driver: Customer-installed: ibm_db under the IBM License Information L-WNKS-F8PR3U for IBM Data Server Driver for ODBC and CLI 12.1.4 (ibm_db itself: Apache-2.0)
Authentication: user and password (secret reference)
Tested environments: IBM Informix Developer Edition (container) 15.0.1.0.3DE
Last verified: 2026-10-13 · connector 0.1.0

Known limitations (7)

Requires agent 0.4.3 or later: agent 0.4.2 and earlier do not contain the Informix connector, so build the IBM extension image FROM that release's digest.
Customer-installed driver: the standard agent image does not include ibm_db. Build the DataNivra agent extension image (agent/runtime/extensions/ibm-db2, shared with IBM Db2), which installs ibm_db and IBM's CLI driver under IBM's licence terms that you accept; without it the agent reports CONNECTOR_NOT_AVAILABLE.
Linux/amd64 agents only: IBM publishes no Linux arm64 CLI driver, so an arm64 agent reports CONNECTOR_PLATFORM_UNSUPPORTED.
Needs a DRDA listener on the server (drsocssl for TLS, drsoctcp without) and a logged database; the SQLI protocol is not used.
A non-ANSI database gives PUBLIC write privileges on every new table, Select on every new sequence and Execute on every new routine; the read-only proof refuses the source until you revoke them (or create objects with NODEFDAC=yes).
User and password authentication only (no Kerberos, PAM challenge or single sign-on).
SET, MULTISET, LIST, ROW and user-defined types are read as their text form.

Details and docs for IBM Informix · Conformance evidence for IBM Informix

Available now Files

Local files (Parquet / CSV)

Shipped in the standard agent and backed by recorded conformance evidence.

Built-in file-set connector for Parquet or CSV extracts in a local or mounted directory (for example a read-only landing zone for exports).

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: filesystem permissions
Tested environments: Local filesystem (Parquet via Apache Arrow) pyarrow 25.0.1
Last verified: 2026-10-08 · connector 0.1.0

Details and docs for Local files (Parquet / CSV) · Conformance evidence for Local files (Parquet / CSV)

Available now Mainframe

Mainframe files (EBCDIC / copybook)

Shipped in the standard agent and backed by recorded conformance evidence.

Built-in connector for mainframe record files described by COBOL copybooks: EBCDIC code pages, fixed and variable-length records, zoned and packed decimal (COMP-3) and binary fields, decoded inside your network into the same masking and certification workflow.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: filesystem permissions
Tested environments: EBCDIC record files (cp037, FB + VB) with COBOL copybooks IBM record formats F/FB/V/VB
Last verified: 2026-10-08 · connector 0.1.0

Known limitations (5)

Verified with synthetic EBCDIC datasets written by DataNivra's own record encoder; not yet verified against unload files produced on a z/OS system.
Reads unloaded extract files (QSAM/VSAM unloads) from a landing directory; it does not connect to z/OS, VSAM or Db2 for z/OS directly.
REDEFINES, OCCURS DEPENDING ON and floating-point (COMP-1/COMP-2) items are refused with a reason code.
Variable-length (V/VB) files are read in one pass up to a configurable size limit.
Files carry no keys: relationships come from industry-pack templates or explicit mappings.

Details and docs for Mainframe files (EBCDIC / copybook) · Conformance evidence for Mainframe files (EBCDIC / copybook)

Available now Databases

MariaDB

Shipped in the standard agent and backed by recorded conformance evidence.

MariaDB through the MySQL connector, with its own compatibility record: MariaDB is tested separately and never assumed compatible because a MySQL driver connects.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: password (secret reference)
Tested environments: MariaDB 11.4.13
Last verified: 2026-10-08 · connector 0.1.0

Known limitations (6)

Requires agent 0.2.0 or later; agent 0.1.0 does not include its driver.
Password authentication only.
Tested against MariaDB 11.4 LTS only.
Roles granted to roles (nested roles) are not expanded; such a source is refused.
System-versioned (temporal) tables are read at their current state only.
The conformance run used TLS without certificate verification against a self-signed test server.

Details and docs for MariaDB · Conformance evidence for MariaDB

Available now Databases

Microsoft SQL Server

Shipped in the standard agent and backed by recorded conformance evidence.

Dedicated SQL Server connector (pure-Python TDS driver): catalog discovery, primary and foreign keys, row estimates from sys.partitions, aggregate profiles and bounded streaming reads into the local engine, over verified TLS by default.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: SQL authentication (secret reference)
Tested environments: Microsoft SQL Server 2022 (Linux container) 16.0.4295.3
Last verified: 2026-10-08 · connector 0.1.0

Known limitations (4)

Requires agent 0.2.0 or later; agent 0.1.0 does not include its driver.
SQL authentication only; Microsoft Entra ID (service principal) and Windows integrated authentication are not supported yet.
Tested against the SQL Server 16.0 engine (Linux container); Azure SQL Database, Azure SQL Managed Instance and Amazon RDS for SQL Server are compatibility profiles still to be tested.
Geography, geometry, hierarchyid, xml and sql_variant columns are carried as text.

Details and docs for Microsoft SQL Server · Conformance evidence for Microsoft SQL Server

Available now NoSQL

MongoDB

Shipped in the standard agent and backed by recorded conformance evidence.

MongoDB connector (pymongo): collection discovery, field shapes inferred from a bounded sample inside your network, _id or unique-index-backed business keys, document counts, aggregate profiles and bounded streaming reads with nested documents carried as JSON text.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: SCRAM-SHA-256 / SCRAM-SHA-1 username and password (secret reference)
Tested environments: MongoDB 7.0.43; MongoDB 8.0.32
Last verified: 2026-10-08 · connector 0.1.0

Known limitations (7)

Requires agent 0.4.0 or later; earlier agent images do not include the pymongo driver.
Fields are inferred from a bounded $sample per collection (sample_size); a later document whose field type differs stops the read with FIELD_TYPE_DRIFT. Fields outside the inferred columns are carried, never dropped, in one JSON column (_extra_fields).
Nested documents and arrays are masked and provisioned as JSON text, not as nested fields.
Only regular collections are read (no views or time-series collections).
SCRAM password authentication only (no X.509, LDAP, Kerberos or AWS IAM authentication).
MongoDB declares no relationships: they come from industry packs or explicit mappings. Business keys other than _id need a unique, non-partial index.
Verified against MongoDB 8.0 (TLS verified) and 7.0 Community in Docker; MongoDB Atlas and Amazon DocumentDB are not yet tested.

Details and docs for MongoDB · Conformance evidence for MongoDB

Available now Databases

MySQL

Shipped in the standard agent and backed by recorded conformance evidence.

Dedicated MySQL connector (PyMySQL): information_schema discovery, primary and foreign keys, row estimates, aggregate profiles and unbuffered streaming reads.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: password (secret reference)
Tested environments: MySQL 8.4.11
Last verified: 2026-10-08 · connector 0.1.0

Known limitations (5)

Requires agent 0.2.0 or later; agent 0.1.0 does not include its driver.
Password authentication only (IAM database authentication not yet tested).
Tested against MySQL 8.4 LTS; Amazon RDS for MySQL and Aurora MySQL are compatibility profiles still to be tested.
Spatial values are read and provisioned as WKT text; BIT values as unsigned integers.
The conformance run used TLS without certificate verification against a self-signed test server; verified TLS against a trusted certificate is not yet part of the recorded evidence.

Details and docs for MySQL · Conformance evidence for MySQL

Available now Databases

Oracle Database

Shipped in the standard agent and backed by recorded conformance evidence.

Dedicated Oracle connector (python-oracledb thin mode, no Oracle Client needed): schema discovery, primary and foreign keys, optimizer row estimates, aggregate profiles and bounded streaming reads inside read-only transactions.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: password (secret reference)
Tested environments: Oracle Database Free (container) 23.26.3.0.0
Last verified: 2026-10-08 · connector 0.1.0

Known limitations (4)

Requires agent 0.2.0 or later; agent 0.1.0 does not include its driver.
Password authentication only. The conformance run used plain TCP (the test listener has no TLS endpoint); it proved that the TCPS default refuses such a listener, but a TCPS session and wallet-based mTLS are not yet part of the run.
Tested against Oracle Database Free (23.26); Oracle 19c and Amazon RDS for Oracle are still to be tested.
XMLTYPE and INTERVAL columns are carried as text, DATE as a timestamp, and NUMBER without declared precision as double precision; object, nested-table and VARRAY columns are not read.

Details and docs for Oracle Database · Conformance evidence for Oracle Database

Available now Files

Parquet files

Shipped in the standard agent and backed by recorded conformance evidence.

Built-in file-set connector for Parquet (or CSV) files on a local or mounted filesystem; each declared file or directory is a table.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: filesystem permissions
Tested environments: Local filesystem (Parquet via Apache Arrow) pyarrow 25.0.1
Last verified: 2026-10-08 · connector 0.1.0

Details and docs for Parquet files · Conformance evidence for Parquet files

Available now Databases

PostgreSQL

Shipped in the standard agent and backed by recorded conformance evidence.

Built-in connector (psycopg 3): catalog discovery, row-count estimates, column profiles, foreign-key relationships and streaming Arrow reads for subsetting.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: password (secret reference)
Tested environments: PostgreSQL 16.15; PostgreSQL 16.2
Last verified: 2026-10-08 · connector 0.1.0

Details and docs for PostgreSQL · Conformance evidence for PostgreSQL

Available now SaaS applications

Salesforce

Shipped in the standard agent and backed by recorded conformance evidence.

Salesforce connector: object discovery through the REST describe API (Id as key, single-target lookups as relationships), SOQL query and queryMore reads with a pinned API version, and the org's daily API budget respected.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: OAuth 2.0 client credentials (secret reference); access token (secret reference)
Tested environments: Salesforce (Developer Edition) API v61.0
Last verified: 2026-10-12 · connector 0.1.0

Known limitations (4)

Requires agent 0.4.0 or later (the Salesforce connector ships in the standard agent image).
Verified against one real Salesforce Developer Edition org (REST API v61.0, OAuth 2.0 client credentials through External Client Apps) with synthetic custom objects; other editions, standard objects and large orgs are untested.
Only configured objects are read; polymorphic lookups are not turned into relationships.
base64 (attachment body) fields are not read.

Details and docs for Salesforce · Conformance evidence for Salesforce

Available now SaaS applications

ServiceNow

Shipped in the standard agent and backed by recorded conformance evidence.

ServiceNow connector: table discovery from sys_db_object and sys_dictionary (inherited fields included, sys_id as key, reference fields as relationships) and Table API reads with a pinned API version.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: basic integration user (secret reference); OAuth 2.0 client credentials (secret reference); access token (secret reference)
Tested environments: ServiceNow (Personal Developer Instance) glide-australia-02-11-2026__patch7-09-17-2026
Last verified: 2026-10-13 · connector 0.1.0

Known limitations (4)

Requires agent 0.4.2 or later: earlier agents cannot prove a ServiceNow integration user read-only on current releases (hidden role names, access-control-shortened pages).
Verified against one real ServiceNow Personal Developer Instance (Australia release, Table API v2, basic integration user) with synthetic custom tables; other releases, OAuth on a real instance and large tables are untested.
The integration user needs read access controls on sys_user_has_role, sys_user_role, sys_db_object, sys_dictionary and every configured table (see the setup steps); table write ACLs are not evaluated, the proof rests on snc_read_only.
Currency fields are read as text, empty strings as nulls; password, image and attachment fields are not read.

Details and docs for ServiceNow · Conformance evidence for ServiceNow

Available now Data warehouses

Snowflake

Shipped in the standard agent and backed by recorded conformance evidence.

Dedicated Snowflake connector: INFORMATION_SCHEMA discovery, declared keys, row counts from table metadata, bounded aggregate profiles and Arrow result reads processed inside your network.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: key-pair (secret reference); OAuth token (secret reference); password (secret reference)
Tested environments: Snowflake service
Last verified: 2026-10-07 · connector 0.1.0

Known limitations (4)

Key-pair authentication verified against a real Snowflake account; OAuth and password authentication are not part of the recorded run.
Requires agent 0.3.0 or later; earlier agent images do not include its driver.
Snowflake does not enforce declared foreign keys, so relationships may need explicit mappings.
NUMBER(p,0) columns are carried as 64-bit integers; larger values fail the job with a reason code.

Details and docs for Snowflake · Conformance evidence for Snowflake

Available now Databases

SQLite (developer evaluation)

Shipped in the standard agent and backed by recorded conformance evidence.

SQLite database files read through the Generic SQL connector with a verified read-only session, for local developer evaluation of the full workflow.

Verified capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: file permissions
Tested environments: SQLite 3.50.4
Last verified: 2026-10-08 · connector 0.1.0

Known limitations (2)

Intended for local developer evaluation, not for production sources.
A SQLite file has no credentials, so there is no separate read-only login to verify; read-only access comes from the file mode and query_only.

Details and docs for SQLite (developer evaluation) · Conformance evidence for SQLite (developer evaluation)

Preview (0)

Real, executable support that you can evaluate today, with the limitations listed on each card. A preview connector has passed the minimum conformance checks on a real engine, not all of them.

No connector currently has this status.

Coming next (7)

Active implementation. These connectors are being built and tested now, but they are not yet available for production use, and the console does not offer them yet.

Coming next NoSQL

Amazon DocumentDB

Active implementation; not yet available for production use.

Amazon DocumentDB (instance-based clusters) through the MongoDB connector, with its own compatibility record: collection discovery, field shapes inferred from a bounded sample inside your network, _id or unique-index-backed business keys, document counts, aggregate profiles and bounded streaming reads. MongoDB compatibility is never assumed: DocumentDB is detected and tested separately.

Planned capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Certification, Evidence generation
Authentication: SCRAM username and password over TLS (secret reference)
Tested environments: None recorded yet
Last verified: Not yet verified

Known limitations (6)

Requires agent 0.4.3 or later: earlier agents treat DocumentDB as MongoDB and refuse it when its proof cannot be completed (fail closed).
DocumentDB clusters are reachable only inside their VPC: the agent must run in that VPC, a peered network or behind a tunnel you operate.
Fields are inferred from a bounded $sample per collection; where a cluster does not support $sample, from the first sample_size documents in natural order.
SCRAM password authentication only (no IAM authentication).
DocumentDB declares no relationships: they come from industry packs or explicit mappings.
Instance-based clusters only; DocumentDB Elastic Clusters are not tested.

Details and docs for Amazon DocumentDB

Coming next Object storage

Google Cloud Storage

Active implementation; not yet available for production use.

File-set connector for Parquet, CSV or JSON-lines objects under a Cloud Storage bucket prefix, read through the Cloud Storage JSON API with bounded listings and ranged reads, into the same discovery, masking, subsetting and certification path as the other object stores.

Planned capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Synthetic workflows, Certification, Provisioning, CI/CD, Evidence generation
Authentication: service-account key (secret reference); workload identity / application default credentials; OAuth access token (secret reference)
Tested environments: None recorded yet
Last verified: Not yet verified

Known limitations (5)

Requires agent 0.4.3 or later: earlier agents do not include the connector (its only driver dependency, google-auth, is already in the standard agent image).
The read-only proof covers IAM on the bucket (including legacy bucket roles); per-object ACLs of fine-grained buckets are not visible to it, so use uniform bucket-level access.
A listing larger than max_list_objects (default 100,000 objects) and objects larger than max_object_bytes (default 10 GiB) are refused rather than read partially.
An access token is not refreshed: it must outlive the job. Service-account keys and workload identity refresh automatically.
Requester-pays buckets and customer-supplied encryption keys are not supported.

Details and docs for Google Cloud Storage

Coming next SaaS applications

Microsoft Dynamics 365

Active implementation; not yet available for production use.

Dynamics 365 (Dataverse) connector: table discovery from Dataverse metadata (primary id as key, single-target lookups as relationships), Web API reads with a pinned version and @odata.nextLink paging.

Planned capabilities: Discovery, Classification, Masking, Deterministic masking, Relationship preservation, Certification, Evidence generation
Authentication: Microsoft Entra ID application, OAuth 2.0 client credentials (secret reference); access token (secret reference)
Tested environments: None recorded yet
Last verified: Not yet verified

Known limitations (4)

Exercised only against a local mock of the Dataverse Web API v9.2; not yet run against a real Dynamics 365 environment.
The application user needs organization-wide Read on every configured table and Read on security roles, users and teams, so the agent can prove it read-only.
Customer and Owner lookups (several target tables) are not turned into relationships; columns with several choices, images, files and column-security-protected columns are not read.
Counts above 5,000 come from Dataverse's snapshot count (RetrieveTotalRecordCount).

Details and docs for Microsoft Dynamics 365

Coming next SaaS applications

SAP applications

Active implementation; not yet available for production use.

SAP S/4HANA connector over OData V2 and V4 APIs: entity sets, keys and referential constraints discovered from each service's $metadata, and counted, key-ordered reads.

Planned capabilities: Discovery, Classification, Masking, Deterministic masking, Relationship preservation, Certification, Evidence generation
Authentication: communication user, basic (secret reference); OAuth 2.0 client credentials (secret reference); access token (secret reference)
Tested environments: None recorded yet
Last verified: Not yet verified

Known limitations (3)

Exercised only against a local mock of SAP Gateway OData V2/V4 services; not yet run against a real SAP system.
Read-only access is attested by the operator; only the services' metadata is verified.
Only single-property referential constraints between configured entity sets become relationships; binary, stream, geo and complex properties are not read.

Details and docs for SAP applications

Coming next Databases

SAP HANA

Active implementation; not yet available for production use.

SAP HANA (Cloud and on-premises) connector using SAP's driver installed by the customer: SYS catalog discovery, primary and foreign keys, M_TABLES row estimates, bounded profiles and reads into the local engine, over verified TLS by default.

Planned capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Certification, Provisioning, Evidence generation
Driver: Customer-installed: hdbcli under the SAP Developer License Agreement (or your SAP agreement)
Authentication: database user and password (secret reference)
Tested environments: None recorded yet
Last verified: Not yet verified

Known limitations (4)

Customer-installed driver: the standard agent image does not include hdbcli. Build the DataNivra agent extension image (agent/runtime/extensions/sap-hana), which installs hdbcli under the SAP Developer License Agreement or your own SAP agreement; without it the agent reports CONNECTOR_NOT_AVAILABLE.
Database user and password only (no X.509, JWT, SAML or Kerberos authentication).
Calculation views with input parameters are not read.
Not yet run against a real SAP HANA system (HANA express needs more memory than the test host has).

Details and docs for SAP HANA

Coming next Data warehouses

Teradata

Active implementation; not yet available for production use.

Teradata Vantage connector using Teradata's driver installed by the customer: restricted dictionary-view discovery, declared primary keys (kept distinct from primary indexes), referential constraints, statistics-based estimates, bounded profiles and reads over verified TLS.

Planned capabilities: Discovery, Classification, Profiling, Masking, Deterministic masking, Entity-aware subsetting, Relationship preservation, Certification, Provisioning, Evidence generation
Driver: Customer-installed: teradatasql under the Teradata License Agreement
Authentication: TD2 user and password (secret reference); LDAP user and password (secret reference)
Tested environments: None recorded yet
Last verified: Not yet verified

Known limitations (4)

Customer-installed driver: the standard agent image does not include teradatasql. Build the DataNivra agent extension image (agent/runtime/extensions/teradata), which installs teradatasql under the Teradata License Agreement that you accept; without it the agent reports CONNECTOR_NOT_AVAILABLE.
A primary index is not a primary key: tables keyed only by a primary index need a declared PRIMARY KEY or a relationship mapping.
Soft referential integrity (WITH NO CHECK OPTION) is reported but not enforced by Teradata.
Not yet run against a real Teradata system (no free local engine).

Details and docs for Teradata

Coming next SaaS applications

Workday

Active implementation; not yet available for production use.

Workday connector: custom reports (Reports-as-a-Service) discovered from their XML schema and REST collections with offset paging, read by an integration system user.

Planned capabilities: Discovery, Classification, Masking, Deterministic masking, Relationship preservation, Certification, Evidence generation
Authentication: integration system user, basic (reports only; secret reference); OAuth 2.0 refresh token of an API client for integrations (secret references); access token (secret reference)
Tested environments: None recorded yet
Last verified: Not yet verified

Known limitations (4)

Exercised only against a local mock of Workday RaaS and REST responses; not yet run against a real Workday tenant (a customer or partner tenant is required).
Read-only access is attested by the operator.
A report is read in one response (no paging) bounded by the configured size limit; incremental extraction is not supported.
Keys and relationships of reports are declared in the source document, not discovered.

Details and docs for Workday

Projected (0)

Strategic roadmap only. Nothing here ships today and no delivery date is committed. Tell us which ones matter to you: demand informs the order.

No connector currently has this status.

Need a source that is not available yet?

Many teams start with an export: land Parquet or CSV extracts in a directory, an S3 bucket or an Azure container inside your network, and read them with a connector that is available today. Each coming-next and projected connector page has a short request form: demand informs the order in which we build, without any promise of a date. You can also read the agent installation guide or learn how customer-resident processing works.

Start free Try the synthetic demo