ROUNVER RESEARCH / SECURITY

Security by design

Security controls, technical scope, and limitations of Regidi Lab Last updated: November 22, 2026.

Security is part of this research's goals. This page explains the controls implemented in the proof of concept and their limitations: a public lab is not the same as an independently audited, certified, production-ready identity service.

  1. 01

    Scope of this policy

    This policy applies to the Regidi Lab website, its static pages, and the demonstration API operated by Rounver Corp. It describes controls in a proof of concept, not contractual security obligations for a commercial platform.

    All scenarios use fictional identities and simulated operations. This version has no real human authentication or KYC/KYB verification and does not demonstrate interoperability with independent issuers or receivers.

  2. 02

    Infrastructure and platform protection

    Regidi Lab uses serverless infrastructure to run the API and deliver static assets. Technical records needed for the demonstration are stored temporarily in a SQLite database. Public site communications use HTTPS.

    Access controls protect administrative infrastructure resources, and additional safeguards are provided by the underlying services. Cloud availability and physical safeguards depend on those services' terms and features; no third-party certification is claimed for Regidi Lab.

  3. 03

    Cryptography and agent identity

    The browser generates a P-256 key pair through Web Crypto. The private key is created as non-exportable and remains in the browser; the corresponding public key is registered for the demonstration.

    The issuer signs delegation credentials with ES256. The simulated receiver verifies both the issuer's signature and the proof that the agent possesses its key, bound to the credential and request details.

    The current proof-of-possession format is specific to Regidi Lab and should not be represented as a complete implementation of a standardized interoperable protocol.

  4. 04

    Delegation, revocation, and replay prevention

    Delegations specify a receiving service, operation, maximum amount, and expiration. Each simulated receiver applies its own policies; a signed permission does not require the receiver to approve an operation.

    The verification process checks revocation status, proof freshness, and replay. Identifiers of consumed proofs are recorded in SQLite to support single-use validation.

  5. 05

    Bot and abuse protection

    Creating a demo agent requires Cloudflare Turnstile verification, performed invisibly when possible. The serverless process validates the token via Siteverify and checks the originating hostname and authorized action before opening a session.

    Request limits are also applied to reduce abuse and excessive consumption. These controls mitigate risks, but do not guarantee detection of all malicious automation.

  6. 06

    Technical data and retention

    We do not request real identities, passwords, banking details, or payment information. SQLite stores session and delegation identifiers, the public part of the agent key, and technical records essential for validating simulated operations.

    Sessions begin with a five-minute lifetime and, after permission is granted, expire with the selected three-, five-, or fifteen-minute delegation. Logical expiry does not necessarily imply immediate physical removal of all records. More details are in the Privacy Policy.

  7. 07

    Vulnerabilities and responsible disclosure

    If you discover a vulnerability, report it to privacy@rounver.com, with a technical description and reproduction steps that do not expose other people's information.

    Please avoid destructive testing, denial-of-service attacks, or disclosure of secrets. We do not currently offer a formal bug bounty or promise financial compensation for reports.

  8. 08

    Incidents, improvements, and limitations

    Following an incident, features may be temporarily suspended, compromised configurations revoked, or fixes published to protect the environment. Legally required notifications will be handled according to the incident's type and scope.

    Regidi Lab is evolving experimental research. Its controls are not a substitute for an independent audit, comprehensive penetration testing, real human identity verification, or a security assessment of a production deployment.