What Entropa's architecture actually does, each claim backed by real code or a public document you can check yourself — not a certification, and not marked done unless it genuinely is.
Every checkpoint is signed with ML-DSA (NIST FIPS-204), independently verified against NIST's own published test vectors — not just "should be correct."
Records are folded into a signed Merkle checkpoint. A later record can never rewrite an earlier one's proof.
Every record page runs a real inclusion-proof and signature check and shows the actual result — not a claim taken on faith.
Entropa never receives your underlying data — only a hash you compute yourself, plus an optional short label you choose to disclose.
When you run the app yourself, your records, your keys and your explorer stay on your Linux machine. We do not hold them.
The records are plain files you can copy, back up and verify offline at any time. We never hold your data, so there is nothing of yours for us to keep or delete.
All traffic to our public sites and demonstration node is HTTPS/TLS.
The node accepts decisions only from callers your platform authenticates, and identity never comes from anything the request body claims.
Deploy, runtime, and admin roles are separate GCP service accounts, each scoped to only the permissions it actually needs.
Keys and credentials live in Secret Manager, never as literals in committed code — enforced by an automated secret-scanning gate on every commit.
Every dependency is checked against known vulnerabilities on every push and on a daily schedule, not just at release time.
The signed, tamper-evident log is the recordkeeping and evidence component that parts of DORA, PCI DSS, NIST SP 800-53 (AU-9(3), AU-10), EU AI Act Articles 12 and 19, and HIPAA 164.312 ask for. Our own estate is assessed against these clauses in a published self-assessment, gaps listed first. This is evidence mapped to those standards, not a full compliance program and not the regulation itself.
Our own demonstration node runs on Google Cloud's own HITRUST-assessed platform, and the application-layer controls Entropa owns directly (least-privilege IAM, no PII in the data path, secrets never logged) are built to that same bar as a default, not an afterthought. Stated honestly: not independently certified — adherence to the controls is the real, verifiable claim.
The terms of use say plainly: provided as is, no warranty or guarantee, use at your own risk, no liability. They are drafted by the author and not yet reviewed by an attorney. See the terms.
There is no uptime commitment, no credits and no guarantee. See no service level.