Security and Compliance Templates¶
Evidence (dated)
This page records observations at a stated revision and date. Re-verify before relying on it; it is not a permanent contract.
Filled 2026-10-09
This page was empty. The inventory below is verified against the repositories. Responsibilities follow canonical ADR-0101 (Authorization Server as sole issuer; Identity Service for accounts; gateway validates tokens; backends enforce authorization; Shell owns frontend auth) and ADR-0100 (tenancy). Factory-specific rules are in tenancy and scope and tool policies.
What exists¶
| Asset | Repository | Role |
|---|---|---|
| Authorization Server template | ConnectSoft.AuthorizationServerTemplate (25 projects) |
OpenIddict-based OAuth2/OIDC issuer; runs in DerechHaLev |
| Identity template | ConnectSoft.IdentityTemplate (24 projects) |
Users, credentials, SMS 2FA, roles; runs in DerechHaLev |
| Identity and Authorization admin MFEs | Blazor.Mfe.Identity.AdminTemplate, .Identity.SelfServiceTemplate, .AuthorizationServer.AdminTemplate |
Administration and self-service surfaces |
| Identity platform and resource server libraries | Extensions.IdentityPlatform, Extensions.ResourceServer, Extensions.Http.OAuth2 |
Token validation, resource protection, OAuth2 clients |
| Web security and limits | Extensions.WebSecurity, Extensions.RateLimiting, Extensions.RequestTimeout |
Headers, rate limits, timeouts |
| Compliance and audit | Extensions.Compliance (classification, redaction), Extensions.AuditNet |
Audit.NET integration; redaction classifiers |
| Tenancy | Extensions.Saas.Abstractions, .AspNetCore, .NHibernate, .Orleans, .Options, .Testing |
Canonical ADR-0100 implementation foundations |
| MCP governance | Extensions.ModelContextProtocol with Microsoft.AgentGovernance |
Policy paths, fail-closed, sanitization, audit for MCP (opt-in) |
Gaps for the Factory¶
No policy engine, approval policy, secret broker, sandbox egress control or audit aggregate; MCP endpoint unauthenticated by default in the agent host; Factory roles and delegation defined by ADR-0015 (2026-10-09) but not implemented; AI-specific threat controls (prompt and tool injection, cross-tenant retrieval, poisoned knowledge) exist only as design (Trust & Operations).