Trust Center · Security

Security posture

How Rightshold is secured — and how to check

Rightshold's security model is unusual in one way: most of it is not a promise, it is a public record. The rules that govern money, grades and access are published as documents whose SHA-256 hashes are sealed into a public, append-only ledger before they take effect. The code that runs the platform reports its own commit and build hash. Anyone can verify the signatures, the ledger and the anchors without an account, without asking us, and without trusting this page.

Policy changes are sealed before they bind

• Every policy that governs fees, grades, arbitration or access is published at https://www.rightshold.com/governance with a content hash sealed in the ledger.
• Rule changes are announced at least 7 days before they take effect (the waiting-period rule, itself sealed). Changes: https://www.rightshold.com/transparency/changes
• A change whose announced hash does not match its sealed hash does not bind anyone. The verification is mechanical: hash the document, compare it with the sealed entry.

Signing and key management

• Every receipt, statement and ledger head is signed with Ed25519. Current and historical signing keys, with validity windows: https://www.rightshold.com/.well-known/rightshold-keys.json
• Keys rotate on a published schedule; retired keys remain published so old signatures stay verifiable.
• Verification never requires our cooperation: the witness kit at https://www.rightshold.com/witness/verify checks signatures, ledger consistency and Bitcoin anchors offline.

Access control

• API access is by per-account key or OAuth 2.1 grant (PKCE required). Grants are scoped per client, listed at GET /v1/me/connections, and revocable per client; every grant, approval, denial and revocation seals a receipt on the ledger.
• Sensitive operations require a fresh step-up confirmation bound to the request fingerprint, valid for five minutes and usable once.
• Accounts carry a freeze control (GET /v1/me/freeze) that halts agent activity immediately.

Testing as a control

• A conformance suite (940+ tests at the last full run) runs on every change; a failing suite does not ship.
• The platform's own public pages carry machine-checked honesty guards: claims made on public surfaces are tested against the live record, and stale or inflated numbers fail the build.
• Conformance evidence: https://www.rightshold.com/conformance

Responsible disclosure

Security reports: security@rightshold.com. Safe harbor: good-faith research that avoids privacy violations, service degradation and data destruction will not be pursued, and we ask for a reasonable window to fix before publication. What to test against, and the rules of engagement: https://www.rightshold.com/security

What we do not claim

No SOC 2, ISO 27001 or similar certification is claimed anywhere on this page, because none is held. Where a control does not exist yet, our pages say so — the ledger's finality status, the witness count and the SLA position are all published as they are, not as we would like them to be.