Readiness

:::info Source Sourced from services/assessment-service/SERVICE_READINESS.md in the documentation repo. :::

1. Level per Milestone

M	Level	Scope
M0	—	Design
M1	L3	Basic MCQ/fill-blank quizzes + scoring
M2	L3	AI question generation (authoring)
M3	L4	Branching scenarios + SCORM 2004 scoring
M4	L4	—
M5	L4	AI rubric grading (high-risk AI); adaptive difficulty

2. Gates

G1 Domain

• QuizBank, BranchingScenario, AttemptResult aggregates + invariants.
• DAG validation; grading rule evaluation.
• 95% unit / 80% mutation.

G2 API

• OpenAPI; idempotency; cursor.
• Pact: consumer from authoring; producer to progress.

G3 Events

• assessment.quiz_bank.*, assessment.attempt_result.scored.v1, assessment.branching_scenario.completed.v1.
• Outbox + inbox.

G4 Sync

• QuizBank + Scenario server_authoritative; AttemptResult append_only.

G5 AI

• AIClient port only.
• Prompts registered + eval sets.
• Rubric grading classified high-risk; HITL enforced.
• Provenance on every AI artifact.

G6 Observability

• Scoring SLIs; AI override rate; appeal rate.

G7 Performance

• Serve p95 < 200ms.
• Scoring p95 < 500ms.
• AI grade p95 < 3s.

G8 Security

• Answer keys encrypted; tenant iso green.
• Anti-cheat controls tested.

3. SLOs

Quiz serve < 200ms p95 · Scoring < 500ms p95 · AI grade < 3s p95 · 99.99% availability.

4. DoD

• Tests green.
• OpenAPI + Pact + event schemas.
• AI prompt eval pass if touched.
• Migration fwd+back.
• Two-tenant iso green.

5. Release Checklists

S1 (M1)

• Basic quizzes functional.
• Offline scoring proven.
• Progress event triggers certification correctly.

S4 (M3)

• Branching scenarios GA.
• SCORM 2004 scoring passes conformance.
• Answer-key encryption audited.

S6 (M5)

• AI rubric grading GA (behind feature flag + HITL).
• Bias monitoring quarterly report.
• Appeal workflow complete.