# STRIDE Checklist — Lilith Platform

**Document Version**: 1.1  
**Last Updated**: 2025-10-03  
**Owners**: Platform Security Guild

This checklist summarises how each Lilith component addresses the six STRIDE
threat categories. Status tags follow this convention:

- ✅ Implemented & continuously monitored
- 🟡 Partially implemented — tracked on the security backlog with delivery dates
- 🔴 Not started — requires immediate attention (none outstanding as of this
  revision)

| Component                                           | Spoofing                                                                         | Tampering                                                                         | Repudiation                                                  | Information Disclosure                                                        | Denial of Service                                                        | Elevation of Privilege                                                |
| --------------------------------------------------- | -------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- | ------------------------------------------------------------ | ----------------------------------------------------------------------------- | ------------------------------------------------------------------------ | --------------------------------------------------------------------- |
| **BFF Service**                                     | ✅ mTLS between services, JWT issuer/audience validation, strict correlation IDs | ✅ Input validation layer, schema coercion, response signing for downstream calls | ✅ Structured request logging with immutable correlation IDs | ✅ Response scrubbing, pii-safe serializers, security headers                 | ✅ Rate limiting, circuit breakers, backpressure aware SSE streaming     | ✅ Persona-level RBAC enforcement, service-to-service scopes          |
| **Authentication Service**                          | ✅ WebAuthn, OAuth device binding, rotated signing keys                          | ✅ Encrypted JWT signing keys in HSM, signature verification                      | ✅ Audit ledger with user/session fingerprinting             | ✅ Bcrypt+pepper, encrypted secrets, zero knowledge proof for password resets | ✅ Adaptive throttling, incremental lockouts, CAPTCHA escalation         | ✅ Role lattice enforced in token claims, privileged path allow-lists |
| **Conversation Service**                            | ✅ Request signing, auth context propagation                                     | ✅ Message digests, append-only histories, integrity checks                       | ✅ Event sourcing with append-only logs and tamper detection | ✅ Field-level encryption for transcripts, scoped retrieval APIs              | ✅ Connection quotas, sliding window limits, fan-out protection          | ✅ Persona isolation, per-conversation ACLs, administrative approvals |
| **AI Orchestration**                                | ✅ Provider certificate pinning, signed provider manifests                       | ✅ Prompt firewall, context sanitisation, deterministic streaming filters         | ✅ Prompt/response hashing with provenance logs              | ✅ Prompt redaction, fine-grained context segmentation, data minimisation     | ✅ Adaptive budget caps, queue backoffs, provider failover mesh          | ✅ Policy engine gating tool-call privileges, trust zone separation   |
| **Content Service**                                 | ✅ Signed ingestion manifests, curator authentication                            | ✅ Version snapshots with hash verification, immutable history                    | ✅ Editorial workflow ledger with reviewer signatures        | ✅ Attribute-based filtering, zero-trust content access                       | ✅ Staged publish pipeline with cache warming guardrails                 | ✅ Moderator/curator RBAC, principle of least privilege               |
| **Media Service**                                   | ✅ Signed URL issuance, token-bound playback                                     | ✅ Integrity checksums, tamper-evident waveform manifests                         | ✅ Delivery receipts with request signing                    | ✅ DRM encryption, least-privilege object policies                            | ✅ Tiered CDN with origin shield, adaptive bitrate protections           | ✅ Admin-only transcoding queues, separation of duties                |
| **Notification Service**                            | ✅ Mutual auth with providers, OAuth client secrets rotation                     | ✅ Template immutable storage, checksum validation                                | ✅ Delivery logs with verifiable receipts                    | ✅ Data minimisation in payloads, encryption at rest                          | ✅ Queue depth guards, exponential backoff, provider circuit breaking    | ✅ Scoped API keys, tenant-aware topic ACLs                           |
| **Data Stores (PostgreSQL, Redis, Object Storage)** | ✅ IAM role separation, certificate auth, Vault-issued creds                     | ✅ Row level security, Crypto signatures for critical tables                      | ✅ WAL shipping with append-only archives, tamper warnings   | ✅ Transparent data encryption, field level tokenisation                      | ✅ Connection pooling limits, query governors, protective indexes        | ✅ DB roles scoped per service, automated rotation                    |
| **Observability / Telemetry Stack**                 | ✅ Authenticated ingestion endpoints, signed agent configs                       | ✅ Immutable log pipeline with sigstore attestations                              | ✅ WORM storage buckets, retained trace IDs                  | ✅ PII scrubbing filters, differential privacy budgets                        | ✅ Multi-tenant rate limiting, log ingestion throttles                   | ✅ RBAC dashboards, break-glass workflow with approvals               |
| **Client Applications (Web & Mobile)**              | ✅ PKCE/OAuth, signed release artefacts, app attestation                         | ✅ Runtime integrity checks, CSP + SRI enforcement                                | ✅ In-app audit journaling, analytics event signing          | ✅ Secure storage APIs, privacy-safe analytics, consent gating                | ✅ Offline request quotas, exponential backoff, offline cache expiration | ✅ RBAC capabilities in-app, hardened admin consoles                  |

## Governance & Continuous Improvement

- **Review cadence**: Quarterly security guild review; critical updates on
  demand
- **Metrics tracked**: Unresolved STRIDE items, time-to-mitigation, number of
  backlog exceptions
- **Tooling**: Threat Dragon for modelling snapshots, Sigstore for artefact
  assurance, Backstage security catalog for control owners

Any change to architecture must re-run this checklist and update artefacts
referenced by `tests/security.threat-model.spec.ts`.
