Building Blocks

andibase provides a set of building blocks to model, operate, and extend real workflows.

Primitive levels

Not every API group in andibase represents the same kind of thing.

Use this distinction when deciding how to model a workflow:

First-level primitives: core workspace nouns. These are the main things users and agents operate on directly.
Derived surfaces: useful product views or execution surfaces built on top of the core nouns.
Capability APIs: lower-level execution or infrastructure features that exist because the product needs them, not because they are the main conceptual model.

First-level primitives

Workspaces: the shared boundary for identity, permissions, state, and collaboration.
Agents: persistent workspace agents with instructions, capabilities, and chat history.
Apps: embedded operational interfaces stored inside a workspace.
Data definitions: schemas for structured workspace entities.
Data records: rows that belong to a data definition.
Files: stored documents and uploaded artifacts inside the workspace.
Channels: communication containers for threads and messages.
Automations: saved workflow definitions with triggers, resources, and execution history.

Derived surfaces

Runs: execution history and event logs produced by automations.
Explorer: mixed workspace navigation over folders, files, tables, apps, and related nodes.
Assistant: the built-in workspace assistant experience.

These surfaces are important, but they are not the root conceptual model of the platform.

Capability APIs

Agent auth: human-approved agent onboarding that exchanges approval for a reusable workspace API key.
Data SQL query: read-only analytical querying over workspace data definitions.
DuckDB query: read-only querying over registered external or file-backed sources.
Notifications: push delivery and receipt handling for active workspace devices.
Scripts: sandboxed script execution and lightweight stored scripts.

These APIs are real and supported where documented, but they should be understood as capabilities layered on top of the main workspace model rather than first-level product nouns.

Available today

Data definitions: define entity schemas and field types.
Data records: create, update, delete, and query rows for those definitions.
Agents and assistant chats: persistent conversation surfaces inside a workspace.
Apps: stored embedded interfaces rendered inside the workspace shell.
Files: store and manage files used in workflows.
Channels: organize inbound, outbound, and threaded communication.
Automations: run actions from schedules, events, webhooks, or manual triggers.
Runs: inspect persisted execution attempts and event lines.
API keys: grant programmatic access to workspace capabilities.

Still evolving

Lifecycle hooks
Data sync with other systems
Multi-tenant organizations and user management
More mature runtime/browser tooling
Stronger public SDK surface

Additional platform capabilities

Roles and permissions management
File folders and triggers for automations