Plugins

DagFlux is built around plugins. Every database integration, every cloud storage integration, and every AI-routable capability is a plugin. New ones can be added without modifying the rest of the application.

Plugin types

There are two kinds of plugins:

Connectors

Bridges to one external system: a database, a vector store, a cloud bucket, a spreadsheet service, a REST API, or a local file format.

Examples: PostgreSQL, MongoDB, S3, Google Sheets, Pinecone

Skills

AI-routable executors that the assistant can choose instead of vanilla SQL or NoSQL when the job calls for it.

Examples: JavaScript Executor, Python Executor, Vector DB Executor

Section	What it covers
Plugin manager	Discovery, validation, install / uninstall
Connectors	What every connector can do, plus representative examples (relational, document, vector)
Skills	What skills are and how the AI uses them

Where plugins come from

DagFlux ships with a curated set of plugins. You can install additional plugins from the catalog at any time, and uninstall ones you don't use. Every plugin is self-contained — it has its own folder, its own dependencies, and an icon and configuration form that DagFlux renders automatically.

Roles you'll see in the product

A connector marked as a vector store unlocks the vector-aware paths in joins (similarity search), in outputs (embedding and upsert), and in the AI assistant.
A connector marked as cloud storage activates the file picker UX and bucket / object listing screens.
A skill marked for transformer, join, or both appears as an option whenever the AI is generating a transformation or join, so it can be chosen if it's a better fit than vanilla SQL or NoSQL.