Skip to main content

Overview

This section covers the primary analysis surface that turns a staged input bundle into a run-scoped risk artifact, report package, thresholds, provenance records, and reproducibility materials.

What the example output demonstrates​

The example_output/ bundle shows the major artifact families produced by the computational safety diligence pipeline:

  • run_manifest.json for run identity and execution metadata
  • plans/ for plan resolution and tier compliance
  • raw/ for module-native outputs
  • results/ for the report-oriented and decision-oriented outputs
  • thresholds/ for category-aware scoring and measurement policy
  • inputs/ for input hashing
  • seal/ for sealing state
  • repro_pack/ for rebuild and verification materials

This section explains how to read those outputs as one coherent system rather than as an unrelated file dump.

What you can do here​

  • How a run is configured and categorized
  • How outputs are grouped and interpreted
  • How evidence state, failure modes, and risk channels are represented
  • How plan resolution and missing-proof handling affect claim strength
  • How reproducibility and verification artifacts fit into the final package
  1. Supported Project Types
  2. Create and Configure a Run
  3. Outputs Overview
  4. Interpret Results