repster 0.1.0

Developer-first replication platform — SDK and CLI

Repster

Lightweight scheduling and observability for dlt pipelines.

dlt loads your data; Repster runs it. Add one decorator to an existing dlt pipeline and get a scheduler, run history, retries, and a web UI. No orchestrator to deploy, no service to babysit. When pipelines need to run without your laptop open, rp deploy moves the same code to Repster Cloud (managed scheduling and an isolated serverless worker per run).

uv add repster

Docs

User documentation is published to docs.repster.online.

Section	Purpose
Quickstart	build and run your first flow
Why Repster?	how it compares to cron, orchestrators, and managed ELT
CLI reference	all rp commands
Configuration	profiles, env vars, pyproject.toml
How-to guides	recipes for specific tasks
Explanation	concepts and background

Local UI

repster dev starts a local web UI at http://localhost:8080: see all discovered flows, trigger runs, and stream logs. No account needed. Deployments, secrets, and team management live in the Repster Cloud UI (repster.online).

Philosophy

Pipelines belong in code, state should be visible and replayable, and pricing should reflect actual compute consumed, nothing more. Pipelines are code and every operation around them is a CLI command, which makes Repster as operable by AI coding agents as by humans. The SDK is open source (Apache-2.0) so developers can trust it: everything the CLI transmits to Repster Cloud is inspectable. The managed layer exists because running reliable distributed infrastructure is genuinely hard, and that's what Repster takes off your plate. The code is yours; we help you run it.

Developing the docs

The docs site is built with MkDocs Material; dependencies are declared in pyproject.toml. From the repster/ directory:

uv run --group docs mkdocs serve   # live-reload server at http://127.0.0.1:8001
uv run --group docs mkdocs build   # static site in site/ (what GitHub Pages deploys)