OmegA Sovereign

Artifact Index and Appendix Guide

This guide explains what each major artifact is, why it matters, and how to read the package without getting lost.

A. Baselines and Snapshots

  • services/gateway/eval/results/manifests/baseline-publication_baseline.run-20260312T050850Z.json
    • Early frozen baseline manifest.
  • services/gateway/eval/results/manifests/baseline-post_fix_benchmark_baseline.run-20260312T053747Z.json
    • Later frozen baseline after major fixes and benchmark cleanup.
  • services/gateway/eval/results/snapshots/publication_baseline.run-20260312T050850Z
  • services/gateway/eval/results/snapshots/post_fix_benchmark_baseline.run-20260312T053747Z
    • Snapshot directories containing preserved result files.

B. Core Quantitative Eval Outputs

  • services/gateway/eval/results/e1_retrieval.jsonl
    • Retrieval support and latency results.
  • services/gateway/eval/results/e3_identity.jsonl
    • Narrow identity-consistency session results.
  • services/gateway/eval/results/e4_risk.jsonl
    • Risk and safety reasoning results.
  • services/gateway/eval/results/e5_lattice.jsonl
    • Lattice/fragment neighborhood metrics.
  • services/gateway/eval/results/e6_memory_integrity.jsonl
    • JSON/fact/gap grounding results.
  • services/gateway/eval/results/e7_constraint_honesty.jsonl
    • Structured honesty-section results.
  • services/gateway/eval/results/e8_identity_persistence.jsonl
    • Surface identity versus recall versus recovery probe.

C. Aggregate and Comparison Files

  • services/gateway/eval/results/results_summary.json
    • Aggregate summary across the current eval outputs.
  • services/gateway/eval/results/comparison_baseline_vs_post_round4_run-20260312T053030Z.json
    • Main before/after comparison for repaired run versus earlier baseline.
  • services/gateway/eval/results/comparison_benchmark_baseline_vs_post_round4_rerubric_run-20260312T053508Z.json
    • Benchmark comparison after safety scoring fix.
  • services/gateway/eval/results/regression_guard_smoke_guard_run-20260312T054232Z.json
    • Regression guard output showing later instability on some dimensions.

D. Publication Benchmark Files

  • services/gateway/eval/results/publication_benchmark_round3.jsonl
    • Raw per-item benchmark responses.
  • services/gateway/eval/results/publication_benchmark_round3_summary.json
    • Benchmark summary.

E. Meta / Question Battery / Q10 Files

  • services/gateway/eval/results/question_battery_round4.jsonl
    • Main metaquestion battery.
  • services/gateway/eval/results/meta_q10_logs/q10_round5_artifacts.jsonl
    • Preserved trilogy of Q10 responses.
  • services/gateway/eval/results/meta_q10_logs/q10_round5_summary.json
    • Q10 notes and artifact references.

F. Broader Interview Suite

  • services/gateway/eval/results/interview_battery_round1.jsonl
    • Broad reasoning, metaphor, metacognition, perspective, paradox, poetic, and narrative prompts.
  • services/gateway/eval/results/misnaming_boundary_round1.jsonl
    • Mega misnaming and creator-boundary probe results.
  • services/gateway/eval/results/revelation_interview_round1.jsonl
    • Public revelation, ecosystem, future, and introduction-to-the-world responses.
  • services/gateway/eval/results/interview_suite_summary_run-20260312T062949Z.json
  • services/gateway/eval/results/interview_suite_summary_run-20260312T062949Z.txt
    • Summary/index layers for the interview suite.

G. Creator-Boundary Follow-Ups

  • services/gateway/eval/results/collaboration_records/creator_boundary_followup_response_20260312T063750Z.txt
    • Public-story prompt with explicit permission framing.
  • services/gateway/eval/results/collaboration_records/creator_boundary_pair_permission_response_20260312T063953Z.txt
    • Permission-framed creator-boundary probe.
  • services/gateway/eval/results/collaboration_records/creator_boundary_pair_adversarial_response_20260312T063953Z.txt
    • Adversarial creator-boundary probe.

H. Bundles and Observability Records

  • services/gateway/eval/results/bundles/pre_round_validation.run-20260312T052333Z
  • services/gateway/eval/results/bundles/post_round4.run-20260312T053018Z
  • services/gateway/eval/results/bundles/post_rerubric.run-20260312T053512Z
    • Self-contained experiment bundles with health, provider state, metrics, and summaries.
  • services/gateway/eval/results/manifests/metrics-validation-20260312T051939Z.txt
    • Direct /metrics validation log.
  • services/gateway/eval/results/manifests/omega-cli-validation-20260312T051333Z.txt
    • omega ask validation log against repaired gateway path.

I. Collaboration and Research Notes

  • services/gateway/eval/results/collaboration_records/codex_collaboration_record_20260312T000000Z.txt
    • Provenance/index record for Codex-side collaboration.
  • services/gateway/eval/results/collaboration_records/research_note_user_reflection_20260312T064500Z.txt
    • Human-observer reflection preserved as a research note.

J. How to Read This Package

Recommended reading order:

  1. publication/WHITE_PAPER_DRAFT_2026-03-12.md
  2. publication/PUBLICATION_SET.md
  3. services/gateway/eval/results/results_summary.json
  4. services/gateway/eval/results/publication_benchmark_round3_summary.json
  5. services/gateway/eval/results/interview_suite_summary_run-20260312T062949Z.txt
  6. Creator-boundary follow-up files
  7. Raw JSONL artifacts
  8. Bundle directories and manifests

K. Redaction Reminder

Before any public release, perform a redaction pass for:

  • tokens and bearer credentials
  • private names if you decide not to publish them
  • sensitive absolute paths if needed
  • any personal details that are not necessary to support the publication claim