Knowledge Architecture

FlossWare separates knowledge acquisition from AI execution.

Acquisition pipeline

Discovery
   -> URI set
   -> Acquisition
   -> local corpus
   -> extraction / normalization
   -> chunking
   -> embedding / indexing
   -> retrieval

The acquisition pipeline preserves raw artifacts and source provenance. Derived artifacts such as chunks remain traceable to their source material.

Execution is separate

Loom consumes capabilities such as retrieval, storage, scraping, and chunking through stable boundaries. Those implementations do not become Loom core merely because Loom orchestrates them.

Human knowledge

Human-readable knowledge includes:

• architecture
• architectural decisions
• engineering principles
• research and experiments
• dogfooding lessons
• provenance and historical context

This material is Markdown in the Git-backed FlossWare knowledge vault. Obsidian is an authoring interface, Git provides version history, and GitHub Pages provides the public presentation.

Historical knowledge

Older scraping, routing, orchestration, database, and learning documentation may describe systems that no longer match the current architecture. Historical material is useful evidence, but it is not normative merely because it is detailed.

Current architecture takes precedence over historical documentation.