Skip to main content

Outputs Overview

The example bundle shows the output package as a layered decision system. The files are grouped by function rather than by audience only.

Output families in the example bundle​

Run identity and planning​

  • run_manifest.json
  • plans/*_module_plan.json

These files explain what the run was, which category it resolved to, and which modules were expected.

Raw module outputs​

  • raw/*_raw.json
  • raw/sar_explorer/*

These are the module-native artifacts and should be treated as the closest layer to the execution source.

Results​

  • results/summary.json
  • results/phase_5_fastfail_summary.json
  • results/phase_5_fastfail_summary.html
  • results/phase_5_computational_safety.html
  • results/phase_5_pipeline_analytics_snapshot.*
  • results/context_of_use.json
  • results/evidence_state_machine.json
  • results/failure_mode_ontology_output_v1.json
  • results/risk_channels_map.json

These files are the main outputs used for interpretation.

Threshold and category artifacts​

  • thresholds/category_scores.json
  • thresholds/measurement_profile.json
  • thresholds/risk_summary.json

These artifacts explain how raw and aggregated signals are turned into risk bands and evidence statuses.

Integrity and replay artifacts​

  • inputs/input_hashes.json
  • seal/preseal.json
  • repro_pack/

These files support hashing, pre-seal provenance, rebuild recipes, and verification of reproduced outputs.

How to read the package​

Start with summary.json, then move to fast-fail, then coverage and evidence-state artifacts, and finally the reproducibility materials if you need audit depth.