mise is design-stage (no application code yet); this states the security posture the design commits to and how to report a vulnerability. The design-level threat model (assets, trust boundaries, STRIDE, residual risks) is THREAT-MODEL.md; penetration testing and per-service STRIDE refinement are deferred to the build phase.
Email danh.software@outlook.com with details and reproduction steps. Please do not open public issues for security reports (Issues/Discussions are not monitored — CONTRIBUTING.md). You'll get an acknowledgement; coordinated disclosure is appreciated. mise is provided under AGPL-3.0 without warranty (see LICENSE).
| Concern | Control | Owning doc |
|---|---|---|
| Tenancy / deployment | single-tenant — one instance per enterprise, in the bank's own GCP project, bank-operated; open source, self-hosted from public source | DELIVERY-MODEL · DECISIONS 9 |
| Access / confidentiality | per-corpus access tiers, RLS at the database (not just UI); cross-corpus joins can't leak rows the caller can't see | DATA-GOVERNANCE §2 |
| AI / agent boundary | serve agent is read-only, permission-gated, tier-scoped; evidence is data, not instructions (prompt-injection safe); no fs/bash | AI-GOVERNANCE §5/§8 |
| Evidence-only invariant | models propose, grounding gates, humans attest — no model asserts compliance | AI-GOVERNANCE §1 |
| Supply chain | SAST · SCA · secrets · container scan; SBOM + cosign signing; signed-image admission | CI-CD §3/§4 |
| Secrets | Secret Manager, WIF (keyless) CI, git-ignored creds + gitleaks | CI-CD §5 · DATA-GOVERNANCE §7 |
| Audit trail | every read, attestation, and model turn is logged and reconstructable | DATA-GOVERNANCE §6 · AI-GOVERNANCE §9 |
| Data retention / PDPD | per-layer retention; PII minimized to role+dept; erasure via the org_role resolver |
DATA-GOVERNANCE §9 |
| Residency (open) | Decree 53/2022 + Decree 13/2023 — region-pinning and cross-border processing | DECISIONS 17 |
| Dependency licensing | permissive-only shipped deps, enforced by a CI license gate | LICENSES |
| Threat model | assets · trust boundaries · STRIDE · residual risks | THREAT-MODEL |