Duckle v0.4.1 is out - DuckDB 1.5.4, in-app updates, Custom SQL for duck sources, and proxy support for REST. Read the release notes
Reference

Component reference

Duckle ships 329 components across six kinds - 104 sources (src.*), 128 transforms (xf.*), 59 sinks (snk.*), 12 data-quality validators (qa.*), 19 control-flow nodes (ctl.*), and 7 code runners (code.*). Every component has Basic, Schema, Preview, Advanced, and Validation tabs and a live preview.

104 sources

Files, lakehouse, databases, warehouses, object storage, streaming, SaaS APIs, NoSQL, vector DBs and more, all under src.*.

128 transforms

Fields, rows, aggregates, joins, windows, strings, dates, JSON, arrays, CDC/SCD, AI and search, all under xf.*.

59 sinks

Files, databases, warehouses, object storage, vector DBs, messaging and comms, all under snk.* with universal write modes.

Looking for a connector-only list grouped by system? See the integrations directory. New to Duckle? Start with install & quickstart.

Sources src.*

Sources read data into a pipeline. Connection-backed sources reuse saved, encrypted connections; file sources accept paths, globs and object-storage URIs.

GroupItems
FilesCSV, TSV, Parquet, JSON, JSONL/NDJSON, Excel .xlsx, YAML, TOML, Fixed-width, XML, Apache Avro
Geospatial filesGeoJSON, Shapefile, GeoPackage, KML, GPX, GML (via spatial extension)
LakehouseApache Iceberg, Delta Lake, DuckLake
Embedded DBsSQLite, DuckDB
Network relationalPostgreSQL, MySQL, MariaDB, CockroachDB, SQL Server, Oracle, ClickHouse
Object storageAmazon S3, Google Cloud Storage, Azure Blob, HTTP(S), MinIO, Cloudflare R2, Backblaze B2
WarehousesMotherDuck, Snowflake, BigQuery, Redshift, Databricks SQL, Azure Synapse, DuckDB Quack
StreamingApache Kafka/Redpanda, NATS JetStream, GCP Pub/Sub, RabbitMQ, AWS Kinesis
SaaS RESTSalesforce, HubSpot, Stripe, Shopify, Notion, Airtable, GitHub, GitLab, Jira, Slack, Zendesk and more, plus generic src.rest / src.graphql
API protocolsOData v4, SOAP/XML
NoSQL & searchMongoDB, Cassandra/ScyllaDB, Elasticsearch/OpenSearch, Redis, CouchDB, DynamoDB
Vector/AI DBspgvector, Qdrant, Weaviate, Milvus
File transferFTP/FTPS, SFTP
MailboxIMAP
Webhook listenerInbound webhook endpoint
Desktop clipboardRead from the system clipboard
Git repoRead files from a Git repository
Date parsing: the CSV/TSV Schema panel supports a per-column strptime Format on Date and Timestamp columns.

Transforms xf.*

Transforms reshape, combine and enrich rows between sources and sinks. The visual Map node (xf.map) joins a main input to up to 3 lookups with per-output expressions and a filter.

GroupItems
FieldsMap (visual mapper joining a main input to up to 3 lookups with per-output expressions + filter), Project/Select, Cast, Rename, Add/Drop/Reorder Column, Coalesce, UUID v4
RowsFilter (with reject port), Distinct, Sample, Top N/Limit, Sort, Skip, Top N per Group, Forward/Backward/Constant Fill
AggregateGroup By, Rollup, Cube, Count, Window Aggregate, Cumulative, Approx Quantile (t-digest), Approx Count Distinct (HyperLogLog)
JoinInner, Left, Right, Full Outer, Cross, Lookup, Semi, Anti, Spatial
Set opsUnion, Union All, Intersect, Except
WindowRow Number, Rank, Dense Rank, Lead, Lag, First Value, Last Value, NTile
StringsRegex Replace/Extract/Match, Split, Concat, Trim, Case, Length, Substring, Format, Hash (md5/sha1/sha256), IP Parse, URL Parse, Text Similarity, Base64, Pad
Date/TimeParse, Format, Extract, Diff/Add, Truncate, Timezone, Time Bin, Now, Epoch
NumericRound, Modulo, Abs, Log, Power, Sqrt, Bucketize, Z-Score, Clamp, Sign
JSON/nestedParse, Stringify, Flatten, JSONPath, Merge, Array Aggregate
ArrayExplode/Unnest, Collect List, Element At, Contains, Distinct, Length
Pivot/shapePivot, Unpivot, Denormalize, Normalize, Transpose
CDC/SCDIncremental Load (watermark), Diff Detect, SCD Type 1, SCD Type 2, Merge/Upsert (with delete propagation), DuckLake CDC reader, Row Hash, Audit Stamp
AI/SearchVector Similarity Search (vss), Full-Text Search BM25 (fts), Embeddings, LLM Transform, Classify, Text Chunker, PII Redact, Semantic Dedupe
GeospatialSpatial Distance, Buffer, Intersects
DebugLog Rows, Assert
dbtxf.dbt runs dbt models (dbt-duckdb / Fusion)
AI transforms: for embeddings, LLM transforms, classification and semantic dedupe, see Duckie AI & AI transforms.

Sinks snk.*

Sinks write pipeline output to a destination. All sinks support the same write modes.

GroupItems
FilesCSV, TSV, Parquet, JSON/JSONL, Excel, XML, Avro, Spatial, Iceberg
DatabasesPostgres, CockroachDB, MySQL, MariaDB, SQLite, DuckDB
WarehousesSnowflake, BigQuery, Redshift, MotherDuck, Databricks, DuckLake, Quack
Object storageS3, GCS, Azure Blob (Parquet/CSV/JSON)
NoSQLMongoDB
Vector DBspgvector, Pinecone, Qdrant, Weaviate, Milvus
MessagingKafka, NATS, Pub/Sub, RabbitMQ
CommsEmail (via SMTP)
File transferFTP/FTPS/SFTP
Write modeBehavior
overwriteReplace the destination contents
appendAdd rows to existing data
truncateEmpty the target, then write
upsertUniversal MERGE with optional delete propagation

Data quality qa.*

Validators split their input: passing rows continue on the main port, while failures route to a reject port for quarantine or re-processing.

GroupItems
ValidatorsNot Null, Range, Regex, Uniqueness, Schema Validate
ProfilingColumn Profile, Describe, Histogram
Cleansing & matchingStandardize, Fuzzy Deduplicate (Jaro-Winkler/Levenshtein), Record Match

Control flow ctl.*

Control-flow nodes orchestrate execution, branching, concurrency and error handling across a pipeline (19 total).

GroupItems
OrchestrationRun Job (call a child pipeline with context vars), Iterate, For Each, Schedule trigger
ConcurrencyParallelize (concurrent in-pipeline branches, auto-detected from CPU cores)
BranchingSwitch (route rows to case outputs), Try/fallback
SignalsWait, Die, Warn, Log, and more

Code code.*

Code runners drop down to custom logic when a visual node is not enough (7 total).

RunnerNotes
Custom SQLcode.sql
SQL templateParameterized SQL
Shellcode.shell
JavaScriptboa engine
WebAssemblywasmi
Python-style routinesScripted row logic
Every node has more: per-node Advanced settings (retry attempts, wait, memory limit) and a Plan tab that shows the generated SQL. For a connector-only directory grouped by system, see integrations.