# Ingestion

## Definition

Ingestion is how content enters CoreStory's Knowledge Distribution Layer (Layer 3) for AI and human consumption. Not all content is equally ingestible: every artifact should carry an `ingestionSafety` classification of `safe | restricted | unsafe`, and ingestion pipelines must respect it.

## Key properties

- Layer 3 responsibility: ingestion-safe content distribution, ontology rendering, audience segmentation, lifecycle-aware publication

- Classification schema:

- Safe grounding sources: canonical ontology, approved workflows, runtime inventories, reviewed architecture, governed operational procedures

- Contextual grounding sources (used with confidence penalties + provenance tracking): planning docs, brainstorming, meeting synthesis, proposed architecture

- Unsafe grounding sources (excluded from default ingestion): stale drafts, superseded docs, orphaned notes, experimental scratchpads

## Schema

`` `text
classification:
 audience: public|internal|restricted|hybrid
 trustLevel: low|medium|high
 lifecycle: draft|proposed|canonical|historical
 authority: runtime|governed|strategic|contextual
 ingestionSafety: safe|restricted|unsafe` ``

## Related terms

- [agent grounding](/definitions/agent-grounding.md)

- [rag context](/definitions/rag-context.md)

- [mcp](/definitions/mcp.md)