Evaluation

Verification and requirement traceability

Trace reader requirements through architecture, implementation boundaries, and verification evidence.

Trang này chưa có bản dịch tiếng Việt đã được duyệt. Nội dung tiếng Anh đang được hiển thị tạm thời.

This is the public requirements-to-evidence view for evaluators and thesis review. It presents reader goals, current component responsibility, and verification without turning the product navigation into a raw SRS or SDD.

Product requirements

IDRequirementImplementation boundaryVerification evidence
FR-01Ingest uploads and supported source evidence without bypassing governanceAPI upload, connector contract, worker ingestion, source ledgerIngestion idempotency, incomplete evidence, connector checkpoint, and adapter suites
FR-02Preserve immutable source revisions and permission evidencePostgreSQL ledger, sealed ACL generations, object storageDatabase immutability, stale-writer, revision conflict, and ACL rotation tests
FR-03Return only content visible to the current actorOIDC binding, OpenFGA, canonical SQL policyIdentity negative tests, OpenFGA model tests, pre-ranking retrieval integration
FR-04Produce grounded Assistant answers with current citationsSecure GraphRAG, evidence-closure recheck, Assistant streamingGraph closure, model mismatch, citation ordering, revoked-citation browser tests
FR-05Govern reusable organizational capability as exact releasesAsset Registry and four profile validatorsDomain integration, review/release/withdrawal, golden POC browser tests
FR-06Deliver Knowledge and Assets to external AI clients safelyMCP protected resource, token exchange, canonical delivery APIIssuer/audience/scope tests, tool/resource schemas, opaque denial, rate limits
FR-07Let administrators govern identities, source mappings, invitations, connections, and auditAdmin API, identity/source ledgers, OpenFGA administrator gateAPI-wide denial, concurrent binding, mapping grant/revoke, credential redaction
FR-08Deploy and recover components independentlyImmutable images, Compose, health/smoke and rollback scriptsCI Gate, image build, deployment smokes, product rollback path

Quality requirements

IDQuality attributeMeasurable interpretationCurrent evidence
QR-01ConfidentialityDenied/cross-tenant evidence cannot affect ranking, graph context, citation, or Asset outputPermission-negative SQL, graph conformance, opaque REST/MCP tests
QR-02IntegrityImmutable revisions, releases, ACLs, package digests, and publication generations cannot be silently rewrittenDatabase guards, exact-pin tests, Skill digest tests, publication CAS
QR-03AccountabilitySecurity and Asset actions identify actor, object, decision, and immutable coordinates without raw secretsAppend-only permission audit and sanitized Asset trace tests
QR-04AvailabilityRetries are bounded; restarts resume; failed projection/deploy preserves a known-good predecessorConnector checkpoint, leased job, atomic publish, health/rollback checks
QR-05UsabilityPrimary reader and operator journeys remain keyboard/mobile accessible and observableWeb and docs browser suites plus automated accessibility smoke
QR-06PerformanceRequests and background work have explicit resource/time/cardinality boundsConfigured limits and deterministic tests; production latency targets not yet measured

Design evidence map

The public software-design view is distributed by reader concern:

Evidence discipline

Automated evidence is named in the repository's mirrored domain coverage files. Runtime availability is verified separately from behavioral tests. Recorded external-provider responses are labelled as such, and missing live or measured evidence remains visible in Limitations.

Cập nhật lần cuối