Kepler is a vulnerability database and lookup store and API currently utilising National Vulnerability Database as data sources; implementing CPE 2.3 tree expressions and version range evaluation in realtime.
Setup
Docker (recommended)
We provide a docker bundle with kepler
, dedicated PostgreSQL database and Ofelia as job scheduler for continuous update
docker compose build
docker compose up
Database migration notes
When the application starts checks for pending database migrations and automatically applies them. Remove the --migrate
option to stop when a pending migration is detected
Build from sources
Alternatively you can build kepler
from sources. To build you need rust
, cargo
and libpg-dev
(or equivalent PostgreSQL library for your Linux distribution)
cargo build --release
Data sources
The system will automatically fetch and import new records every 3 hours if you use our bundle, while historical data must be imported manually.
Kepler currently supports two data sources, National Vulnerability Database and NPM Advisories. You can import the data sources historically as follows.
NIST Data
To import NIST records from all available years (2002 to 2022):
for year in $(seq 2002 2022); do
docker run --rm -v $(pwd)/data:/data \
-e DATABASE_URL=postgres://kepler:kepler@localhost:5432/kepler \
--network=kepler_default \
kepler:dev import_nist $year -d /data;
done
The system will automatically fetch and import new records records every 3 hours.
APIs
There are two primary APIs as of right now — the product
API and the cve
API detailed below.
Products API
Products can be listed:
curl http://localhost:8000/products
Grouped by vendor:
curl http://localhost:8000/products/by_vendor
Or searched:
curl http://localhost:8000/products/search/iphone
CVEs API
To use the vulnerabilities search API via cURL (prepend node-
to the product name in order to search for NPM specific packages):
curl \
--header "Content-Type: application/json" \
--request POST \
--data '{"product":"libxml2","version":"2.9.10"}' \
http://localhost:8000/cve/search
Responses are cached in memory with a LRU limit of 4096 elements.