Executive summary
cunningham-chain-data is the publication surface for roots produced by the 2026 first-kind Cunningham-chain campaign. The git repository stays small: it contains a schema, manifest, thousand-row sample, and conversion script. Full cumulative snapshots are attached to versioned GitHub releases. The latest documented snapshot has 1,551,489 roots from CC10 through CC18 and is normalized as CSV. Keeping the dataset separate from the search engine makes release identity, citation, and downstream analysis clearer.
Problem and context
Campaign code and campaign data change at different rates. Storing a multi-million row aggregate in the source history would make the engine repository harder to clone and would blur which snapshot supports a result. This repository gives the data an explicit schema and release lifecycle of its own.
What was published
| Artifact | Purpose |
|---|---|
schema/README.md | Raw and normalized field definitions |
manifest/dataset_manifest.json | Snapshot identity and aggregate counts |
sample/ | A 1,000-root raw and CSV example |
scripts/make_snapshot_csv.py | Deterministic normalization from the raw format |
| GitHub release asset | The full compressed cumulative CSV |
Evidence and dataset shape
The March 19, 2026 snapshot records the following aggregate distribution:
| Range | Count |
|---|---|
| CC10 | 1,262,271 |
| CC11–CC15 | 289,154 |
| CC16 | 55 |
| CC17 | 7 |
| CC18 | 2 |
| Total | 1,551,489 |
The documented bit range is 64–153, with the main campaign concentrated at 87–90 bits. The release is cumulative, not a delta from the earlier snapshot.
Boundaries
The repository publishes campaign output; it does not independently prove the mathematical or priority claims surrounding each root. Those methods and validation tools live in the search-engine repository. Full data is not present in a plain clone and must be obtained from the named release. The earlier raw snapshot remains a separate historical artifact.
Repository map and first use
git clone https://github.com/nmicic/cunningham-chain-data
python3 scripts/make_snapshot_csv.py --helpRead schema/README.md before consuming either sample or release data. Use the manifest to pin a snapshot in analysis work, and follow the release link in the source README for the full compressed CSV.