Provided Job Plugins

BlueCurve provides connectors for the 25 source systems below. Each reference lists accepted configuration keys, types, defaults, validation requirements, and downloadable TOML samples.

Configuration and credentials

Select the installed connector package when creating a job. Set configurationLanguage to Toml and submit the completed TOML as the configuration string.

Sample credentials and identifiers are placeholders. Resolve values from your secret manager before submission and protect the resulting job configuration. The connector parsers do not expand ${ENVIRONMENT_VARIABLE} placeholders or resolve secret_reference automatically. A secret reference does not replace a required password, token, client secret, or certificate path.

File paths refer to the job's execution environment. Coordinate mounts, certificate files, and source connectivity with your administrator. File System root paths have their explicitly documented environment-variable expansion.

Connector reference

Connector	Configuration samples
Aerospike	1
Alfresco	1
Amazon S3	4
ArangoDB	1
Azure Cosmos DB	1
Box	3
Cassandra	1
Confluence	2
Couchbase	1
CouchDB	1
Database	10
DynamoDB	1
File System	3
Google Workspace	4
InfluxDB	1
Jira	2
Microsoft 365	3
MongoDB	2
OrientDB	1
RavenDB	1
Salesforce	4
SAP	3
ServiceNow	4
SurrealDB	1
TigerBeetle	1

Shared behavior

• collection.id takes precedence over indexing.collection_id. Both are optional collection ULIDs.
• scope.source_id takes precedence over indexing.source_id. Supply valid non-empty ULIDs when setting either value.
• sync.delete_missing takes precedence over its delete_detection alias. Defaults differ by connector; check its table before enabling reconciliation.
• Arrays such as [[drives]] or [[tables]] define the actual crawl sources. Similarly named scope keys may be accepted without affecting execution.
• Unknown keys are rejected. Keys marked Ignored are accepted but are not read into the connector's active configuration.
• Accepted resilience and some indexing or sync keys currently have no configurable effect. Their presence does not establish retries, timeouts, or permission ingestion.

Use the connector's exact option table instead of assuming every connector supports every shared key. Start with a limited source scope and confirm document writes and checkpoints before enabling a broad scheduled crawl.

Next: Jobs API