System Atlas
SystemAtlas

Example queries

The same questions over REST, MCP, the CLI and the bulk files with DuckDB.

In short

One question, four ways: the largest dental offices by employees. REST and MCP read the live registry a page at a time; the CLI pages for you; the bulk files answer offline, in one query.

REST

curl -s "https://systematlas.org/api/v1/top?metric=employees&industry=naics_6212&limit=10" \
  -H "Authorization: Bearer sa_…"

MCP

Ask your connected agent, or call the tool directly:

{ "name": "top", "arguments": { "metric": "employees", "industry": "naics_6212", "limit": 10 } }

CLI

systatlas top --metric employees --industry naics_6212 --limit 10
# the files themselves, checked against the manifest's sha256
systatlas download organizations
systatlas download industry:naics_6212 --dataset values --format jsonl
systatlas download source:gleif --dataset organization_links

Bulk files with DuckDB

DuckDB reads the gzipped files as they are, no unpacking. Figures are in values, keyed by key; join to organizations on organization_id.

SELECT o.name, o.region, TRY_CAST(v.value AS DOUBLE) AS employees, v.as_of, v.source_url
FROM 'values.csv.gz' v
JOIN 'organizations.csv.gz' o ON o.id = v.organization_id
WHERE v.key = 'employees' AND starts_with(o.industry_key, 'naics_6212')  -- the node and its children
ORDER BY employees DESC NULLS LAST
LIMIT 10;

Python

import pandas as pd
orgs = pd.read_csv("organizations.csv.gz")
ids = pd.read_json("identifiers.jsonl.gz", lines=True)
orgs.merge(ids[ids.scheme == "lei"], left_on="id", right_on="organization_id").head()
PreviousSources catalogNext Data changelog