Lineage And Provenance

Lineage and provenance explain where Canon knowledge came from, how it changed, and what consumed it later.

What Lineage Records

Lineage should make a packet traceable through:

The purpose is not bureaucracy. It is the ability to answer, months later, why a claim exists and whether it can still be trusted.

Why Provenance Matters

AI-assisted work can look authoritative even when it is based on weak context. Provenance keeps the source visible.

Provenance helps answer:

• who or what authored the input?
• which files or observations grounded the packet?
• which assistant-generated content was later reviewed?
• which claims are evidence-backed?
• which claims are assumptions?
• what changed between draft and publication?

Without provenance, project memory becomes an accumulation of polished but unaccountable text.

Claim Traceability

A useful packet lets readers trace important claims back to evidence.

For example:

• a security finding should point to config, code, dependency data, or threat context
• an architecture decision should point to constraints, alternatives, and consequences
• a domain invariant should point to domain evidence, workflows, or existing behavior
• a verification conclusion should point to exact checks or artifacts

If a claim cannot be traced, keep it as an assumption or open question instead of promoting it.

Promotion Auditability

When packet content is promoted into project memory, it must remain auditable.

Promotion should preserve:

• source packet
• source document
• evidence summary
• approval state
• promotion timestamp or event
• compatibility notes
• known limitations

Promoted content should not become detached from its origin. The promoted memory is more useful when it can still answer why it exists.

Downstream Consumption

Downstream systems should store packet refs and metadata, not just copied prose. Stable refs make it possible to refresh, verify, or reject knowledge later.

For machine-facing integration, use Canon's adapter response fields and authority metadata rather than scraping human summaries.