Telemetry and Read Models

This page describes how Veritas output should be interpreted over time without turning the product into a heavy observability platform.

Canonical Records

The canonical Veritas records are repo-local generated evidence and feedback artifacts:

These artifacts support Standards Feedback.

Derived Summaries

Some outputs are read models rather than source-of-truth records.

Current examples:

Use Readiness Report, Repo Conformance, or Standards Feedback depending on the summary.

Human-Facing Read Model

The intended interpretation path is:

  1. readiness report for a change
  2. repo conformance for standing requirements
  3. standards feedback for trends
  4. optional console or telemetry exports

Questions a read-model layer should answer:

Optional OTLP Export

OTLP is optional. Use it for team dashboards, cross-repo aggregation, or trend analysis in existing observability infrastructure.

Do not use OTLP as the first or only source of truth. Keep rich detail in repo-local generated evidence.

Safe dimensions are small and bounded:

Avoid high-cardinality labels:

Product Guidance

Veritas should stay artifact-first:

  1. generated evidence
  2. readiness and conformance summaries
  3. standards feedback
  4. optional telemetry export
  5. optional console