Archivista is a graph and storage service for in-toto attestations. Archivista enables the discovery and retrieval of attestations for software artifacts.

• Store and retrieve in-toto attestations
• Query for relationships between attestations via a GraphQL API
• Validate Witness policy without the need to manually list expected attestations

• It creates a graph of supply chain metadata while storing attestations that can be later used for policy validation and flexible querying.
• It is designed to be horizontally scaleable, supporting storing a large number of attestations.
• It supports deployment on major cloud service and infrastructure providers, making it a versatile and flexible solution for securing software supply chains.
• It only stores signed attestations to further enhance security and and increase trust.

• Native support for storing attestations created by Witness
• A GraphQL API endpoint and playground
• Support for MySQL and Postgres database backends
• Support for S3-compatible object storage
• A Helm Chart for deployment in Kubernetes environments
• The ability to download and export attestations to transfer across an air gap
• Support for Darwin, Windows, and ARM architectures.

When an attestation is uploaded to Archivista it will store the entire attestation in a configured object store as well as scrape some data from the attestation and store it in a queryable metadata store. This metadata is exposed through a GraphQL API. This enables queries such as finding all attestations related to an artifact with a specified hash or finding all attestations that recorded the use of a specific dependency.

Archivista uses Subjects on the in-toto Statement as edges on this graph. Producers of attestations (such as Witness can use these subjects as a way to expose relationships between attestations.

For example when attesting that an artifact was compiled the compiled artifact may be a subject, as well as the git commit hash the artifact was built from. This would allow traversing the graph by the commit hash to find other relevant attestations such as those describing code reviews, testing, and scanning that happened on that git commit.

A public instance of Archivista is running here for testing purposes. The data in this instance is open to the world and there are currently no SLAs defined for this instance.

Archivista requires a MySQL database as well as a compatible file store. Compatible file stores include a local directory or any S3 compatible store.

A docker compose file is included in the repository that will run a local instance of Archivista along with the necessary services for it to operate. These include Minio and MySQL. Simply cloning the repo and running

docker compose up --build -d

is enough to get a local instance of Archivista up and running. Archivista will be listening at http://localhost:8082 by default with this docker compose file.

Archivista is configured through environment variables currently.

Note: If ARCHIVISTA_ENABLE_SQL_STORE is set to false no metadata about store attestations will be collected. Archivista will only store and retrieve attestations by Gitoid from it's storage. Archivista servers with GraphQL or SQL store disabled cannot be used to verify Witness policies.

Archivista exposes two HTTP endpoints to upload or download attestations:

POST /upload - Uploads an attestation to Archivista. The attestation is to be in the request's body

GET /download/:gitoid: - Downloads an attestation with provided gitoid from Archivista

Additionally Archivista exposes a GraphQL API. By default the GraphQL playground is enabled and available at root.

archivistactl is a CLI tool in this repository that is available to interact with an Archivista instance. archivistactl is capable of uploading and downloading attestations as well as doing some basic queries such as finding all attestations with a specified subject and retrieving all subjects for a specified attestation.

As previously mentioned, Archivista offers a GraphQL API that enables users to discover attestations. When Archivista ingests an attestation some metadata will be stored into the SQL metadata store. This metadata is exposed through the GraphQL API. Archivista uses Relay connections for querying and pagination.

Here is an entity relationship diagram of the metadata that is currently available.

erDiagram
dsse ||--|| statement : Contains
statement ||--o{ subject : has
subject ||--|{ subjectDigest : has
statement ||--o| attestationCollection : contains
attestationCollection ||--|{ attestation : contains
dsse ||--|{ payloadDigest : has
dsse ||--|{ signature : has
signature ||--o{ timestamp : has

dsse {
    string gitoidSha256
    string payloadType
}

statement {
    string predicate
}

subject {
    string name
}

subjectDigest {
    string algorithm
    string value
}

attestationCollection {
    string name
}

attestation {
    string type
}

payloadDigest {
    string algorithm
    string value
}

signature {
    string keyID
    string signature
}

timestamp {
    string type
    time timestamp
}

Archivista can be easily deployed thru the provided helm chart into your kubernetes cluster. See the README for more details.

We would like to expand the types of data Archivista can ingest as well as expand the metadata Archivista collected about ingested data. If you have ideas or use cases for Archivista, feel free to contact us or create an issue!

See CONTRIBUTING.md for information on how to contribute to Archivista.