Deploy the open-source engine in your cloud. Use our Cloud Control Plane to manage it at scale.
Get started
skipprd is a complete, self-serve engine. Install it, point it at Postgres, and land Apache Iceberg tables in your own bucket. The Cloud Control Plane is there when you want fleet management — not as a gate on the software.
Install skipprdDeveloper Docs →
Mid-market VP track: Book a 20-min call · 20 minutes, no sales pitch.
$ brew install skipprd/tap/skipprd$ skipprd connect data-source postgres --pipeline orders --name production$ skipprd connect data-sink skippr-lake --pipeline orders --name lake --warehouse 's3://my-lake/' --catalog-table skippr$ skipprd sync --pipeline ordersA warehouse copy of last night's extract is not enough. Skippr ingests continuously from OLTP systems, SaaS APIs, files and streams, then lets you query the table as it fills. Iceberg snapshot commits are ACID — atomic, so readers never see a half-written table. That is a stream lake: live operational state on open files you own.
Most stacks still batch into a vendor warehouse, then hope another tool can read the same rows. Skippr lands Apache Iceberg tables on Parquet in your object store. Compute is decoupled: skipprd query, Athena, Snowflake, DuckDB, notebooks — the same tables. No second copy. No lock-in.
Why Skippr
Privacy by default. Each source declares append, merge, or replace-by-day — and Skippr refuses to start a pipeline the destination cannot honour.
| Against | Skippr |
|---|---|
| vs. Fivetran | No volume-based pricing anxiety. Data stays inside your VPC. |
| vs. Apache Flink | No complex JVM cluster management. One ultra-lightweight Rust binary. |
| vs. DuckDB | Handles transactional write-locks, micro-inserts, and live schema drift continuously so DuckDB can query cleanly. |
skipprd query runs SQL against the lake table plus rows that just arrived. --watch follows a sync.You keep the systems you already run. Skippr is the stream lake they write to.
AI agents can't reason over yesterday's stale warehouse batches. Skippr feeds live transactional state directly into local Parquet/Iceberg tables, providing a secure, real-time context and memory layer that agents can query locally via DuckDB without data leaking outside your private cloud.
That is the point of a stream lake for AI:
skipprd query, or your own process. Row-level data does not leave the VPC to "give the model context."Read the Skippr ReAct suite, or start from Developer Docs and land the tables first.
Install skipprd, then land Postgres into Iceberg on a bucket you own. Full flags, Python, and every connector live in Developer Docs.
brew install skipprd/tap/skipprd
skipprd --version
skipprd connect data-source postgres --pipeline orders --name production
skipprd connect data-sink skippr-lake --pipeline orders --name lake \
--warehouse 's3://my-lake/' --catalog-table skippr
skipprd discover --pipeline orders
skipprd sync --pipeline orderscurl -sL https://raw.githubusercontent.com/skipprd/skipprd/main/install.sh | sh
skipprd --versionpip install skippr
python -c "import skippr; print(skippr.Session)"connect, discover, sync, query, serve.import skippr.skipprd works completely out of the box. You do not need a call to land Iceberg tables in your own bucket.
Enterprise Jumpstart is optional: we deploy the binary in your VPC, wire the first sources, and hand you the keys. It is a premium onboarding track for teams that want a founder-engineer in the room for the first pipelines — not how Skippr ships, and not a substitute for the software.
Skippr is for ambitious data teams and startups. You might already have a data team, or want a trusted lake fast. The skipprd Rust binary drops into your cloud or on-prem. Data does not leave your deployment.
“Skippr has saved us hundreds of hours of engineering time and thousands of dollars every month. It's a no brainer.”
Works without a callYou own the Iceberg tablesskipprd is open sourceOptional Enterprise Jumpstart
Xero
MES
Margin after scrap
8.4%
Business insight
SKU-14 still makes 8.4%.
Deploy skipprd in your VPC. Use the Cloud Control Plane when you run it at scale.
Integration
Accounts, ERP, CRM, ads, product databases and event streams land in the warehouse or lake you already pay for. Batch loads, change data capture and SaaS APIs run through one engine.
Each source declares how it must be written — append events, merge current state, or rewrite a revised day — and Skippr refuses to start a pipeline the destination cannot honour.
You own the Iceberg tables. The engine is open source and self-hosted — no vendor lock-in. It works out of the box. Enterprise Jumpstart is optional if you want help landing the first pipelines.