Tapstate docs

Learn the tapstate product model, prepare connectors, and build reviewable data paths

Tapstate connects operational systems through one governed Capture–Transform–Serve data path. Use these docs to understand the model, prepare external systems, and build a data path and check its results. Start in the web application; use the CLI, REST API, or MCP when you need programmatic control.

Start here

What is tapstate?

Understand the customer problem and the tapstate product model.

Quickstart

Run the disposable Docker playground and inspect one assembled order.

Connections and connectors

Understand connector definitions, connection roles, and scope.

Capture modes

Choose Snapshot, Full load + CDC, Stream, or API by outcome.

Try the current preview

The Quickstart demonstrates one operational data path:

MySQL orders and PostgreSQL shipments, each captured by an initial load plus continuous change data capture (CDC)
an in-flight nest transform that assembles one order document
continuous materialization to MongoDB

The local MongoDB instance stores the demo's managed view. In Cloud, the output uses the Atlas connection configured for that Cluster. See State and storage for the storage boundaries. Start with the Quickstart to run the sample locally, then build a pipeline for your own data.

The server-backed preview is supplied through the disposable Docker playground; this release does not publish a standalone server installer.

Complete a task

Build a pipeline

Connect a source, save a draft, preview data, and start a pipeline in the web application.

Prepare a connector

Find roles, modes, permissions, limitations, and exact field names.

Explore the order-state demo

Change MySQL and PostgreSQL rows and watch the same MongoDB document update.

Look up fields

Use the DSL grammar and resource reference for exact syntax.

Troubleshoot

Separate authoring, connector, deployment, and data problems.

Understand the product

Architecture

Review the logical data plane, control plane, and state boundaries.

Use cases

Match customer outcomes to the data path they need.

Compare approaches

Choose between a unified path and a composable streaming stack.

State and storage

Understand working, recovery, and materialized state.

Developers and AI agents

Use the CLI for local authoring and scripted operations, the REST API for integrations, and MCP for an AI host. These are optional clients of the same server. Create a workspace when you need reviewable .tap.yml files in version control.

AI-readable documentation

The HTML pages, llms.txt, combined context, and page-level Markdown are generated from the same content. Connector frontmatter supports discovery; the page body remains canonical for people and agents.