Files
Jyotisha/docs/research/worktree_fragment_triage_2026_06_28.md
T
2026-07-08 17:03:04 +08:00

3.0 KiB

Worktree Fragment Triage - 2026-06-28

Purpose

This note records the current dirty-worktree fragments after the high-value bridge passes. It prevents future agents from repeatedly rediscovering the same files and mistaking scratch output for source truth.

Promoted In This Pass

  • Shadbala absolute oracle comparison:
    • <repo>/scripts/shadbala_oracle_comparison.py
    • <repo>/docs/research/shadbala_absolute_oracle_comparison_audit_2026_06_28.md
  • VedAstro range scan allowlist and signal metadata:
    • <repo>/scripts/vedastro_service_adapter.py
    • <repo>/docs/research/vedastro_range_scan_allowlist_audit_2026_06_28.md
  • Marriage benchmark summary bridge:
    • <repo>/scripts/marriage_benchmark_summary.py
    • <repo>/docs/research/marriage_benchmark_summary_bridge_audit_2026_06_28.md

Remaining Fragments

Generated or Scratch Output

Do not promote without a specific new use case:

  • <repo>/full_chart_data.json
  • <repo>/test_dasha.json
  • <repo>/test_output.json
  • <repo>/scratch_extract.py
  • <repo>/scratch_mcp_eval.py

Duplicate Benchmark Output

  • <repo>/tests/verify-results-v6.1.json

The canonical tracked source is:

  • <repo>/tests/test-data/verify-results-v6.1.json

The untracked copy differs mainly by list ordering inside target arrays and does not need promotion as a second source of truth.

Generated Report Candidate

  • <repo>/tests/印度占星实战案例综合验证报告-v6.1-2026-05-03.md

Much of its durable content is already represented by:

  • <repo>/references/verified-patterns-marriage-timing-v6.md
  • <repo>/scripts/marriage_benchmark_summary.py

Only promote this report later if a stable human-facing archive is needed.

Weak Test Candidates

  • <repo>/tests/test_dasha_raman_truth.py
    • Current state: pytest.xfail placeholder with mocked computation.
    • Disposition: useful boundary values were promoted into <repo>/references/oracle/dasha_shadbala_oracle_cases.json as template_bv_raman_vimshottari_boundary_series.
    • Follow-up: collect source artifact metadata before promoting beyond template_only.
  • <repo>/tests/test_yoga_benchmark_cases.py
    • Disposition: promoted into a real benchmark with concrete rule_id expectations and metadata contract checks.
    • See <repo>/docs/research/yoga_benchmark_case_promotion_audit_2026_06_28.md.

Superseded or Needs Review

  • <repo>/scripts/oracle_functional_benefics.py
    • Superseded in strict evidence by the functional benefic/malefic layer in mcp_server.py.
    • Could be promoted later only if rewritten as a JSON CLI around the canonical helper.
  • <repo>/scripts/patch_api_tz.py
  • <repo>/scripts/patch_engine_tz.py
  • <repo>/scripts/sync_to_workbuddy.sh
    • Review manually before promotion. Do not batch-commit as-is.

Benign Generated Timestamp

  • <repo>/references/oracle/artifacts/pyjhora_oracle_artifact_manifest.json

Current diff only changes generated_at. Do not commit timestamp-only churn unless the artifact inventory materially changes.