Ownership Map¶

extended-data owns the base data layer. Surfaces outside that boundary were moved into repositories where their dependencies, docs, tests, and release cadence are first-class.

In This Package¶

Surface	Current owner
Pure data functions	`extended_data.primitives`
Generic and shape-specific extended containers	`extended_data.containers`
File import/export and codecs	`extended_data.io`
Workflow composition and local file sync primitives	`extended_data.workflows`
Input loading and decorators	`extended_data.inputs`
Structured lifecycle logging	`extended_data.logging`

Moved Out¶

Prior surface	Current repository	Install target
External vendor API clients	`jbcom/vendor-fabric`	`vendor-fabric`
Vendor capability dispatch and provider SDK adapters	`jbcom/vendor-fabric`	`vendor-fabric[...]`
Meshy, Slack, Google, GitHub, AWS, Vault, Zoom, Anthropic, Cursor integrations	`jbcom/vendor-fabric`	`vendor-fabric[...]`
Vendor-backed Python sync capabilities	`jbcom/vendor-fabric`	`vendor-fabric[secrets-sync]`
SecretSync agent tool wrappers	`jbcom/agentic-fabric`	`agentic-fabric[...]`
Agent framework integrations	`jbcom/agentic-fabric`	`agentic-fabric[...]`

The old in-package connector and secrets namespaces are intentionally absent. That is a clean major-version boundary: code should depend on the package that owns the capability it uses.

Dependency Direction¶

The intended layering is dependency-inward:

extended-data
  <- vendor-fabric
       <- agentic-fabric

extended-data has no dependency on the higher layers. Higher layers may use extended-data primitives, containers, input handling, workflows, and logging without reimplementing those base concerns. agentic-fabric depends on vendor-fabric for provider coordination; vendor-fabric depends on extended-data for generic data behavior.