DELIVERY: Public Data Reconstruction + ETC Provenance Map

1. Open ETC-public-data-3D.html in a modern browser. It is self-contained and works offline.
   Drag in any direction to rotate, use the wheel to zoom, click a node to focus
   on its neighbors, and use Back to return to the complete graph. No API key is needed.
   Arrow keys also rotate when the graph has focus. Escape returns to the overview.
   Source layers have distinct depths with fixed branch offsets. The Fit current
   view button restores framing without changing the viewing angle. Graph JSON
   includes the fixed 3D coordinates; depth does not add new evidence relationships.

2. public-data-reconstruction.zip contains a general-purpose, English-only skill.
   It has no ETC, country, industry, model, or personal-project default.
   Extract the public-data-reconstruction folder into your agent's skills directory.
   For Codex, use ~/.codex/skills/ and invoke $public-data-reconstruction.
   Start with the data you want; later say "continue" to resume the saved project ledger.

3. ETC-source-to-field-ledger.csv records source-to-component connections.
   ETC-evidence-index.csv records the supporting project-relative files and hashes.
   ETC-source-map.json preserves the editable graph structure.
   ETC-source-contribution.csv explains every source's circle-size class.
   Source circle sizes show documented contribution classes, not percentages or
   accepted-cell counts. Small hollow circles mark definitions, historical work
   and unresolved routes; a small circle does not mean low scientific importance.
   The Chinese method summary and English ETC case study describe the reconstruction.

Scope: This is a source-to-variable-group reconstruction, not a full independent
re-audit of every underlying data cell. Historical candidate batches and current
manuscript sources are distinguished. The source links include discovery portals;
exact supporting records remain in the project audit trail.

The skill ZIP contains only reusable instructions and the ledger utility.
Private conversations, credentials, large raw data, and personal project state
are excluded from this delivery bundle. The ETC graph and method case describe
the research project; review those separately before sharing research material.

Tests: Skill structure and English-only/general-purpose checks passed. The ledger
utility passed synthetic integrity tests. These do not establish that every
public-data search will succeed or that AI will make no factual errors.
