Risk Management Policy

Risk identification, assessment, treatment, and monitoring framework.

NIST controls:RA-1 through RA-5
Last updated:2026-03-15
Category:Governance
# Risk Management Policy **Document ID**: KD-POL-006 **Owner**: Chief Technology Officer **Approved By**: CEO, Arivaran **Effective Date**: 2026-03-19 **Next Review Date**: 2027-03-19 (recomputed 2026-10-02 on the stated semi-annual cycle from the Effective Date; the 2026-09-19 occurrence was not recorded, and this date does not assert that a review has taken place) **Review Cycle**: Semi-annual **Classification**: Internal ## 1. Purpose This policy defines how Arivaran identifies, assesses, treats, and monitors information security risks to KuiperDesk. Risk management is the mechanism that connects business objectives to security controls. Without a structured risk process, controls are either excessive (wasting resources) or insufficient (creating blind spots). This policy ensures that every security investment is traceable to a specific risk and that residual risks are explicitly accepted, not ignored. ## 2. Scope This policy covers information security risks to: - KuiperDesk platform services and infrastructure - Customer data processed, stored, and transmitted by KuiperDesk - Arivaran's ability to deliver backup and restore services per contractual SLAs - Supply chain dependencies and third-party services - Regulatory compliance obligations (GDPR, SOC 2, ISO 27001) ## 3. Policy Statements ### 3.1 Risk Assessment Methodology 3.1.1. Arivaran uses a semi-quantitative risk assessment methodology. Risks are scored on two dimensions: **likelihood** and **impact**, each on a 1-5 scale. **Likelihood Scale**: | Score | Label | Definition | |-------|-------|-----------| | 1 | Rare | Less than once per 5 years, or requires nation-state capability | | 2 | Unlikely | Once per 1-5 years, requires significant attacker skill | | 3 | Possible | Once per year, exploitable by a motivated attacker | | 4 | Likely | Multiple times per year, common attack vector | | 5 | Almost Certain | Expected to occur, or has already occurred | **Impact Scale**: | Score | Label | Definition | |-------|-------|-----------| | 1 | Negligible | No customer impact, internal inconvenience only | | 2 | Minor | Single-tenant degraded service, no data loss, resolved within RTO | | 3 | Moderate | Multi-tenant service disruption, or minor data exposure, regulatory notification not required | | 4 | Major | Data breach affecting multiple tenants, regulatory notification required, customer trust damaged | | 5 | Critical | Complete platform compromise, widespread data loss/exposure, existential business risk | 3.1.2. **Risk Score**: Likelihood x Impact. Risk levels: | Score Range | Level | Treatment Requirement | |-------------|-------|-----------------------| | 1-4 | Low | Accept or monitor. Document rationale. | | 5-9 | Medium | Treat within 90 days or formally accept with compensating controls. | | 10-15 | High | Treat within 30 days. CTO approval required for acceptance. | | 16-25 | Critical | Treat immediately. CEO approval required for acceptance. Cannot be accepted for more than 30 days. | ### 3.2 Risk Identification 3.2.1. Risk identification sources: - **Threat modeling**: Conducted for each major system component (aaagent, aaHashSvc, aaTwinSvc, aaDash) and for infrastructure (RKE2 cluster, PKI chain, network overlay). Updated when architecture changes. - **Vulnerability scanning**: Trivy container scans (CI/CD + weekly), OS vulnerability feeds (AlmaLinux OVAL), dependency audits (vcpkg audit for C++ services, npm audit for aaDash). - **Incident post-mortems**: Every P1-P3 incident generates risk register entries for identified root causes and systemic weaknesses. - **Audit findings**: Internal and external audit findings are mapped to risk register entries. - **Industry intelligence**: CVE feeds for deployed components (PgEdge, CNPG, Keycloak, SeaweedFS, RKE2, Cilium). - **Regulatory changes**: GDPR guidance updates, SOC 2 criteria revisions, ISO 27001 amendment tracking. 3.2.2. Any Arivaran team member may submit a risk for assessment by creating a Forgejo issue with the `risk` label. ### 3.3 Risk Register 3.3.1. The risk register is the authoritative inventory of all identified risks. It is maintained as a structured document in the Forgejo repository and reviewed quarterly. 3.3.2. Each risk register entry contains: ``` Risk ID: RISK-NNN Title: [Short description] Description: [Detailed description of the threat scenario] Affected Assets: [Systems, data, or processes at risk] Threat Source: [External attacker, insider, natural disaster, vendor, etc.] Likelihood: [1-5 with justification] Impact: [1-5 with justification] Inherent Risk Score: [L x I] Current Controls: [Existing controls that mitigate this risk] Residual Likelihood: [1-5 after current controls] Residual Impact: [1-5 after current controls] Residual Risk Score: [L x I] Treatment Decision: [Treat / Accept / Transfer / Avoid] Treatment Plan: [If Treat: specific actions, owner, deadline] Risk Owner: [Person accountable] Last Reviewed: [Date] ``` 3.3.3. The risk register shall contain, at minimum, risks covering: - Supply chain compromise of container images (PgEdge, CNPG, Keycloak, SeaweedFS) - Customer data exfiltration via compromised agent certificate - RKE2 cluster compromise via container escape - Single point of failure in SeaweedFS (001 replication) - OVH provider failure (all infrastructure on one provider) - PKI chain compromise (OpenBao intermediate CA key leak) - Insider threat (small team, broad access) - GDPR enforcement action due to data handling non-compliance ### 3.4 Risk Treatment 3.4.1. **Treat**: Implement additional controls to reduce likelihood or impact. Treatment plans must specify: the control to implement, the responsible person, the deadline, and how effectiveness will be measured. 3.4.2. **Accept**: Formally acknowledge the residual risk without additional controls. Acceptance requires documented justification and approval at the appropriate level (see 3.1.2). Accepted risks are re-evaluated at each quarterly review. 3.4.3. **Transfer**: Shift the financial impact to a third party (e.g., cyber insurance). The underlying risk still requires monitoring; only the financial consequence is transferred. 3.4.4. **Avoid**: Eliminate the risk by removing the activity or asset. Example: not storing customer PII beyond what is required for service delivery. ### 3.5 Risk Review Cadence 3.5.1. **Quarterly Risk Review**: The CTO conducts a formal review of the full risk register quarterly. The review assesses: - Whether risk scores remain accurate given current threat intelligence - Whether treatment plans are on track - Whether accepted risks are still within tolerance - New risks identified since the last review 3.5.2. **Event-Driven Review**: The risk register is updated immediately when: - A P1 or P2 incident occurs - A critical vulnerability is disclosed in a deployed component - Infrastructure architecture changes materially (e.g., new storage backend, new cloud provider) - A new regulatory requirement affects KuiperDesk 3.5.3. **Annual Comprehensive Assessment**: Once per year, a full risk assessment is conducted, including updated threat modeling for all major components. This assessment produces an updated Statement of Applicability (ISO 27001) and feeds into the SOC 2 risk assessment narrative. ### 3.6 Risk Appetite 3.6.1. Arivaran's risk appetite for KuiperDesk: - **Zero tolerance**: Customer data confidentiality breaches, PKI chain compromise, insider data theft - **Low tolerance**: Service availability below SLA commitments, regulatory non-compliance - **Moderate tolerance**: Operational inefficiency, technical debt, non-critical system outages - **Accepted**: Business risks inherent to a startup operating with a small team (e.g., key person dependency) 3.6.2. The risk appetite is reviewed annually by the CEO and CTO and communicated to all personnel. ## 4. Roles and Responsibilities | Role | Responsibility | |------|---------------| | CEO | Define risk appetite. Approve Critical risk acceptance. Allocate resources for risk treatment. | | CTO | Own the risk register. Chair quarterly risk reviews. Approve High risk acceptance. Conduct threat modeling. | | Infrastructure Engineers | Identify technical risks. Implement treatment plans. Report risk events. | | All Personnel | Report identified risks and near-misses. | ## 5. Compliance Mapping | Policy Statement | SOC 2 Criteria | ISO 27001:2022 | |-----------------|----------------|----------------| | 3.1 Assessment Methodology | CC3.1 | Clause 6.1.2 | | 3.2 Risk Identification | CC3.2 | Clause 6.1.2 | | 3.3 Risk Register | CC3.2, CC3.3 | Clause 6.1.2, 6.1.3 | | 3.4 Risk Treatment | CC3.4 | Clause 6.1.3 | | 3.5 Review Cadence | CC3.1 | Clause 8.2, 9.3 | | 3.6 Risk Appetite | CC3.1 | Clause 5.2 | ## 6. Exceptions There are no exceptions to maintaining the risk register or conducting quarterly reviews. Individual risk acceptance decisions follow the approval levels defined in Section 3.1.2. ## 7. Enforcement Failure to conduct scheduled risk reviews is escalated to the CEO. Undocumented risk acceptance (taking a known risk without formal registration) is a policy violation reportable under KD-POL-003. ## 8. Revision History | Version | Date | Author | Changes | |---------|------|--------|---------| | 1.0 | 2026-03-19 | CTO | Initial release |

This document is part of the Arivaran Twin compliance program. For questions or the latest version, contact compliance@arivaran.ai.

Release-ready. Saved on this browser.