CommandAGI · commandagi.com/for/datasets

CommandAGI
Open the workbench
Menu
Open the workbenchSign in

Datasets

Public data, every row pinned.

Each dataset's published stage is imported into the Drive, one table a kind, every row pinned to its stage and its raw input. Keepers grow the series every day.

Open the data appRead the formats

Authored in git, produced in the Drive

Declared once, grown by machines.

A person or an agent declares a series. A keeper reads it from its source's API and appends to its CSV. An import pins every row to its input.

How a dataset reaches the DriveA series is declared in demand.csv in git. The statistics keeper reads its source's API on a timer and appends new values to the series' CSV in the Drive, with the source URL, the hash of the answer and the time read. The world-model import stages each dataset, uploads one table a kind, pins each row to its stage and raw input, and writes the index the graph reads.statistics/demand.csva series, declared in gitThe keeperdaily: BTS, UNHCR, FRED, BLS …The world-model importstaged, resumable, pacedstatistics/<topic>/<scheme>/<ser-ies>.csvgrows onlyindexes/world-model/the graph's indexHow a dataset reaches the DriveA series is declared in demand.csv in git. The statistics keeper reads its source's API on a timer and appends new values to the series' CSV in the Drive, with the source URL, the hash of the answer and the time read. The world-model import stages each dataset, uploads one table a kind, pins each row to its stage and raw input, and writes the index the graph reads.statistics/demand.csva series, declared in gitThe keeperdaily: BTS, UNHCR,FRED, BLS …The world-modelimportstaged, resumable,pacedstatistics/<topic->/<scheme>/<s-eries>.csvgrows onlyindexes/world--model/the graph's index

Read the formats

A series says where each value came from.

US airline passengers a month, from the Bureau of Transportation Statistics, as the keeper wrote them in CommandAGI's Drive. The source URL and the answer's SHA-256 are on every row; shortened here.

statistics/transport/bts/passengers-total.csvreal rows · dev · read 2026-11-04 09:53 UTC
bts:passengers-total  iso3:USA  2026-M06  87,468,000 passengers  observed 2026-11-04T09:53:24Z
bts:passengers-total  iso3:USA  2026-M05  84,693,000 passengers  observed 2026-11-04T09:53:24Z
bts:passengers-total  iso3:USA  2026-M04  80,400,000 passengers  observed 2026-11-04T09:53:24Z
source   https://data.bts.gov/resource/crem-w557.json?$select=date,u_s_airline_traffic_total_non_seasonally_adjusted…
evidence sha256:a77293a7…  reader program:scripts/statistics/keep.mjs adapter:bts-t100

Counted in the Drive

13 public datasets, 14.5 million rows indexed.

The public manifest of the world-model index on dev, 2026-11-04: 806 tables and 2,827,242 subjects (case-studies/world-model/stats.json).

Series that grow. 2,763 series files in 77 schemes under statistics/ on dev, 35.5 GB, among them UNHCR's displacement totals and BTS's passengers. Capital faucets. Grants.gov, SBIR, CHIPS, DOE loans, the EU's programmes and more: their openings, closings and filings are 353 event files under events/capital/ on dev. Rights are kept. Each source's terms are inventoried. A dataset that may not be shared is stored private.

public datasets indexed
13
rows
14,497,202
subjects
2,827,242
series files, 35.5 GB
2,763
Indexed rows of the largest public datasets, thousands
The numbers
Indexed rows of the largest public datasets, thousands
DatasetThousand rows
ACS 5-year4488
USGS earthquakes3528
GLEIF parents1901
Treasury debt1267
V-Dem951
USDA PSD919
ProvenanceFile cloud://home/user-commandagi/indexes/world-model/public.csv (dev)Output case-studies/world-model/stats.jsonTool scripts/sites/case-studies.mjs world-modelRan on This computer (dell-gb10-1), reading the dev DriveWhen

Bring your own data.

Put a CSV in a folder of your Drive and open it in the workbench.

Open the data appSee the graph explorer