Verifiable Archive

​

What it is

​

What you get

• Evidence-grade integrity. A document’s hash is bound into a post-quantum-signed entry; the chain proves the whole archive’s order and completeness. Verification needs only the public key.
• Encryption with one method, everywhere. ML-KEM-1024 + AES-256-GCM (a fresh per-document key) on local disk or S3 alike — the storage target is a plug-in, the cryptography is not.
• Storage you choose. A local-filesystem backend for on-prem and air-gapped deployments, or an S3 / S3-compatible backend (Amazon S3, MinIO, Cloudflare R2, Wasabi) for durable object storage — same evidence model either way.
• Reporting and auditing that proves continuity. A read-only auditor produces signed audit reports and continuous integrity attestations — chained “still intact as of T” proofs of custody — plus retention reports and regulator-verifiable evidence packs.
• Feeds the tools you already run. Audit events and metrics export to SIEM (JSON-lines, syslog, CEF), Prometheus/Grafana (/metrics), and GRC/BI (CSV) — it owns the cryptography and integrates with your dashboards, rather than replacing them.
• Optionally Substrate 2-bound. Fold archive entries and audit reports into the broader Substrate 2 evidence chain.

​

How it works

​

Packages

​

Backups

import hashlib
from algovoi_doc_archive import Archive, LocalFsBackend, HybridPqcEncryptor, signing
from algovoi_doc_archive.encryption import generate_mlkem_keypair

# One-time keys: Falcon-1024 signs entries; ML-KEM-1024 encrypts documents.
pk, sk         = signing.generate_keypair()
kem_pk, kem_sk = generate_mlkem_keypair()        # keep kem_sk offline — needed only to restore

# Public ML-KEM key only → the backup host can sign + encrypt, but cannot read past backups.
arc = Archive(secret_key=sk, public_key=pk,
              storage=LocalFsBackend("/srv/archive"),
              encryptor=HybridPqcEncryptor(kem_pk))

backup = open("db-2026-06-09.sql.gz", "rb").read()

# Hash-only: record existence + integrity, no extra storage.
e = await arc.notarize(doc_hash="sha256:" + hashlib.sha256(backup).hexdigest(),
                       title="db-2026-06-09", metadata={"host": "db01"})

# ...or encrypt-and-store the bytes (post-quantum encryption at rest):
# e = await arc.notarize(data=backup, title="db-2026-06-09", content_type="application/gzip")

assert (await arc.verify_chain()).valid          # whole-chain integrity + ordering — offline

• Write-only nodes — give the backup host only the ML-KEM public key; the secret key lives in an offline vault and is needed only to retrieve() and restore.
• Two modes — hash-only is a signed, ordered, tamper-evident ledger of your existing backups with no extra storage; encrypt-and-store also keeps the bytes encrypted at rest (ML-KEM-1024 + AES-256-GCM).
• Independent audit — anyone holding the Falcon public key can verify the entire chain offline, with no AlgoVoi service and no secret. Swap LocalFsBackend for the S3 backend for object storage, same chain.

​

Cryptography

• Post-quantum on both axes. Entries are signed with Falcon-1024 (NIST Level 5); documents are encrypted with ML-KEM-1024 + AES-256-GCM. Both the integrity (signatures) and the confidentiality (encryption) are post-quantum — not just the signatures.
• The append-only chain is SHA-256-linked (quantum-resistant) and verifies offline.
• Canonical form verified across 8 implementations. Entry preimages use RFC 8785 (JCS) canonicalization — the discipline in AlgoVoi’s open conformance corpus. A real archive entry canonicalized byte-identically across eight independent JCS implementations (Python, Node.js, Ruby, PHP, Go, Rust, Java, .NET), so the exact bytes a signature covers are reproducible in your own stack. The 8-language guarantee is the canonical form; the Falcon-1024 signature is verified with the published Python reference.

​

Why commercial

• Embed without attribution overhead. A commercial OEM licence lets you ship it inside your own product with no open-source notice obligations.
• Enterprise terms. Support, warranty, and a defined relationship — for a records-critical dependency.
• Maintained crypto suite. Post-quantum primitives and long-horizon record formats are ongoing work, bundled into the licence.

​

Who it’s for

• Regulated record-keeping — broker-dealers, investment firms, and any organisation with tamper-evident, long-horizon retention obligations.
• Legal, IP, and contracts — proof a document or invention existed on a date, unaltered.
• Healthcare and public sector — long-lived immutable records, including air-gapped on-prem.
• Anyone who needs records that an auditor or regulator can verify independently, without trusting the storage vendor.

​

Get Verifiable Archive