FAC-98 feat: add audit trail MVP by y4nder · Pull Request #215 · CtrlAltElite-Devs/api.faculytics

y4nder · 2026-03-29T21:04:20Z

Summary

Add append-only AuditLog entity backed by a BullMQ AUDIT queue with dual emission paths (interceptor + direct emit)
Tag 11 MVP endpoints across auth, moodle sync, questionnaires, and analysis modules
Include full documentation: architecture doc, 4 decision records, workflow updates

Changes

New Files (14)

File	Purpose
`src/entities/audit-log.entity.ts`	Append-only entity (no CustomBaseEntity, no FK to User)
`src/modules/audit/audit.module.ts`	`@Global()` module — cross-cutting concern
`src/modules/audit/audit.service.ts`	Fire-and-forget `Emit()` with error swallowing
`src/modules/audit/audit.processor.ts`	BullMQ processor, concurrency: 1, direct DB write
`src/modules/audit/audit-action.enum.ts`	12 MVP action constants
`src/modules/audit/dto/audit-job-message.dto.ts`	Queue envelope interface
`src/modules/audit/dto/emit-params.dto.ts`	Emit parameter interface
`src/modules/audit/decorators/audited.decorator.ts`	`@Audited()` SetMetadata decorator
`src/modules/audit/interceptors/audit.interceptor.ts`	RxJS `tap()` post-response emit
`src/modules/audit/testing/audit-test.helpers.ts`	Shared test utilities for controller specs
`src/modules/audit/audit.service.spec.ts`	Service unit tests
`src/modules/audit/audit.processor.spec.ts`	Processor unit tests
`src/modules/audit/interceptors/audit.interceptor.spec.ts`	Interceptor unit tests
`docs/architecture/audit-trail.md`	Full architecture documentation

Modified Files (18)

src/configurations/common/queue-names.ts — Added AUDIT
src/entities/index.entity.ts — Registered AuditLog
src/modules/index.module.ts — Added AuditModule to ApplicationModules
src/modules/auth/auth.controller.ts — Tagged logout with @Audited()
src/modules/auth/auth.service.ts — Refactored Login/RefreshToken with direct audit emits outside transactions
src/modules/moodle/controllers/moodle-sync.controller.ts — Tagged sync trigger + schedule update
src/modules/questionnaires/questionnaire.controller.ts — Tagged submit, ingest, wipe
src/modules/analysis/analysis.controller.ts — Tagged pipeline create/confirm/cancel
src/modules/auth/auth.service.spec.ts — 7 new audit tests + @optional test
src/modules/moodle/controllers/moodle-sync.controller.spec.ts — Added interceptor mocks
src/modules/questionnaires/questionnaire.controller.spec.ts — Added interceptor mocks
src/modules/analysis/analysis.controller.spec.ts — Added interceptor mocks
docs/architecture/core-components.md — Added AuditModule, AUDIT queue
docs/decisions/decisions.md — Added decisions feat: implement automated releases and fix test reporting #34-37
docs/workflows/auth-hydration.md — Added audit events section

Key Design Decisions

No FK to User — actorId is a plain string so audit records survive user deletion
@Global() module — only application module to use this; avoids friction in adopting @Audited()
Dual paths — interceptor for standard endpoints, direct emit for auth events lacking CLS context
Sanitized metadata — login failure reasons use fixed codes, never raw error.message
Metadata cap — interceptor limits captured params/query to 4KB

Test plan

All 622 existing + new tests passing
ESLint clean
New tests: AuditService (4), AuditProcessor (5), AuditInterceptor (9), AuthService audit (7)
Adversarial review completed (15 findings, 9 fixed)
Verify migration applies cleanly on fresh database
Manual smoke test: login → check audit_log table for auth.login.success row

Closes #214

🤖 Generated with Claude Code

Initialize tech-spec for audit trail feature covering auth events, admin actions, and sensitive data mutations via BullMQ queue. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

Add detailed file references, code patterns, MVP endpoint targets, and testing strategy from Step 2 codebase investigation. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

Add 20 tasks across 5 phases, 11 acceptance criteria, testing strategy, and risk analysis. Step 3 of quick-spec workflow complete. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

Rename WIP to tech-spec-audit-trail-mvp.md and mark status as ready-for-dev. Quick-spec workflow complete (steps 1-4). https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

@optional

Add 6 preventions from pre-mortem analysis: @optional() AuditService injection in AuthService, Logger.warn on failed emissions, attempts:1 on audit jobs, CLS null-safety warnings, bulk ingestion documentation, and full payload logging on processor failures. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

@optional

Log full emission params on failure, add @optional() undefined test case for auth service, add CLS null-metadata warning test for interceptor. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

@global

Critical: @global() AuditModule, explicit interceptor stacks on all MVP endpoints, fix DI scoping for host modules. High: move login/refresh to direct emit, extend @Audited() with resource option, add AuditAction enum, tap vs finalize docs. Medium: drop repo stub, fix occurredAt, add attempts:1 in code, separate browserName/os fields, fix role claim, complete file list. Low: clarify emit location, entity array registration, test Redis, interceptor ordering, endpoint count. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

Fix 14 findings: emit audit AFTER transactions with void (fire-and- forget), replace existing @UseInterceptors instead of adding duplicates, AuditInterceptor falls back to JWT payload (no CurrentUserInterceptor required), import AppClsModule instead of CommonModule, PII-safe logging in failed handler, PostgreSQL DEFAULT on occurred_at, fix AC 4 syntax and Task 6 Zod contradiction, document empty metadata for /ingest and null resourceId for created-by-action scenarios. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

@optional

Critical: add explicit transaction refactoring pattern with code snippets for Login() and RefreshToken() — emit after transaction returns, not inside. High: correct @optional() resilience claim, fix resourceType for wipe endpoint, add per-endpoint interceptor bullets, acknowledge logout MetaDataInterceptor gap. Medium: add metadata schema table, full UUID regex, AuthenticatedRequest type, justify @global() convention, full FileInterceptor config, fix /ingest CurrentUserInterceptor contradiction. Low: create()+flush() pattern, processor log counter, fix test description, base class name. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

Fix metadata table accuracy (4 rows showed wrong shapes), add null- safe destructure for RequestMetadataService in direct emit, strategy Name fallback for minification, type action as AuditAction end-to-end, consistent PII redaction policy, add CurrentUserInterceptor to POST /submissions (highest-volume), make occurredAt required (no JS default), document MetaDataInterceptor DI resolution, clarify attempts config, note create-action resourceId gap, Redis failed job monitoring guidance, actorId is plain string not @manytoone, logging side effect note. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

@Audited

Implement append-only AuditLog entity backed by a BullMQ AUDIT queue with dual emission paths — interceptor for authenticated endpoints and direct emit for auth events where CLS context is unavailable. - AuditLog entity (no CustomBaseEntity, no soft delete, no FK to User) - AuditService with fire-and-forget Emit() that never breaks requests - @Audited() decorator + AuditInterceptor (RxJS tap, post-response) - @global() AuditModule — cross-cutting concern, no explicit imports - 11 MVP endpoints tagged: auth (login/logout/refresh), moodle sync, questionnaire (submit/ingest/wipe), analysis (create/confirm/cancel) - Sanitized failure metadata (fixed reason codes, no raw error.message) - Metadata size cap (4KB) on interceptor-captured params/query - actorId index for "what did user X do?" queries - Shared audit test helpers to DRY controller spec boilerplate - 622 tests passing, lint clean

@optional

* docs: add audit trail MVP tech-spec WIP Initialize tech-spec for audit trail feature covering auth events, admin actions, and sensitive data mutations via BullMQ queue. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: update audit trail tech-spec with deep investigation findings Add detailed file references, code patterns, MVP endpoint targets, and testing strategy from Step 2 codebase investigation. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: generate full implementation plan for audit trail tech-spec Add 20 tasks across 5 phases, 11 acceptance criteria, testing strategy, and risk analysis. Step 3 of quick-spec workflow complete. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: finalize audit trail MVP tech-spec as ready-for-dev Rename WIP to tech-spec-audit-trail-mvp.md and mark status as ready-for-dev. Quick-spec workflow complete (steps 1-4). https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply pre-mortem hardening to audit trail tech-spec Add 6 preventions from pre-mortem analysis: @optional() AuditService injection in AuthService, Logger.warn on failed emissions, attempts:1 on audit jobs, CLS null-safety warnings, bulk ingestion documentation, and full payload logging on processor failures. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: add team refinements to audit trail tech-spec Log full emission params on failure, add @optional() undefined test case for auth service, add CLS null-metadata warning test for interceptor. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply all 23 adversarial review fixes to audit trail spec Critical: @global() AuditModule, explicit interceptor stacks on all MVP endpoints, fix DI scoping for host modules. High: move login/refresh to direct emit, extend @Audited() with resource option, add AuditAction enum, tap vs finalize docs. Medium: drop repo stub, fix occurredAt, add attempts:1 in code, separate browserName/os fields, fix role claim, complete file list. Low: clarify emit location, entity array registration, test Redis, interceptor ordering, endpoint count. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 2 adversarial review fixes to audit trail spec Fix 14 findings: emit audit AFTER transactions with void (fire-and- forget), replace existing @UseInterceptors instead of adding duplicates, AuditInterceptor falls back to JWT payload (no CurrentUserInterceptor required), import AppClsModule instead of CommonModule, PII-safe logging in failed handler, PostgreSQL DEFAULT on occurred_at, fix AC 4 syntax and Task 6 Zod contradiction, document empty metadata for /ingest and null resourceId for created-by-action scenarios. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 3 adversarial review fixes to audit trail spec Critical: add explicit transaction refactoring pattern with code snippets for Login() and RefreshToken() — emit after transaction returns, not inside. High: correct @optional() resilience claim, fix resourceType for wipe endpoint, add per-endpoint interceptor bullets, acknowledge logout MetaDataInterceptor gap. Medium: add metadata schema table, full UUID regex, AuthenticatedRequest type, justify @global() convention, full FileInterceptor config, fix /ingest CurrentUserInterceptor contradiction. Low: create()+flush() pattern, processor log counter, fix test description, base class name. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 4 adversarial review fixes to audit trail spec Fix metadata table accuracy (4 rows showed wrong shapes), add null- safe destructure for RequestMetadataService in direct emit, strategy Name fallback for minification, type action as AuditAction end-to-end, consistent PII redaction policy, add CurrentUserInterceptor to POST /submissions (highest-volume), make occurredAt required (no JS default), document MetaDataInterceptor DI resolution, clarify attempts config, note create-action resourceId gap, Redis failed job monitoring guidance, actorId is plain string not @manytoone, logging side effect note. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * FAC-98 feat: add audit trail MVP Implement append-only AuditLog entity backed by a BullMQ AUDIT queue with dual emission paths — interceptor for authenticated endpoints and direct emit for auth events where CLS context is unavailable. - AuditLog entity (no CustomBaseEntity, no soft delete, no FK to User) - AuditService with fire-and-forget Emit() that never breaks requests - @Audited() decorator + AuditInterceptor (RxJS tap, post-response) - @global() AuditModule — cross-cutting concern, no explicit imports - 11 MVP endpoints tagged: auth (login/logout/refresh), moodle sync, questionnaire (submit/ingest/wipe), analysis (create/confirm/cancel) - Sanitized failure metadata (fixed reason codes, no raw error.message) - Metadata size cap (4KB) on interceptor-captured params/query - actorId index for "what did user X do?" queries - Shared audit test helpers to DRY controller spec boilerplate - 622 tests passing, lint clean --------- Co-authored-by: Claude <noreply@anthropic.com>

@isnotempty

… Trail MVP (#217) * chore: add /audit skill with auto GitHub issue creation * FAC-87 FAC-88 fix: async bcrypt and expired token filtering in RefreshToken #199 Replace bcrypt.compareSync with concurrent async bcrypt.compare via Promise.all to avoid blocking the event loop (#195). Add expiresAt filter to the DB query so expired tokens are excluded before running expensive bcrypt comparisons (#197). Closes #195 Closes #197 https://claude.ai/code/session_01XttX7NLr6HY7TDLddDk2AG Co-authored-by: Claude <noreply@anthropic.com> * FAC-89 fix: username logged in plaintext in Moodle login error path #200 Closes #198 * FAC-90 fix: LoginRequest DTO missing @isnotempty and length constraints #201 Empty strings and arbitrarily long payloads were passing validation on the login endpoint. Add @isnotempty() and @maxlength() decorators to both username and password fields, consistent with RefreshTokenRequestBody. Add DTO validation tests covering empty and oversized inputs. Closes #196 * FAC-91 FAC-92 fix: include teacher role in faculty lookup and use EnrollmentRole enum #206 getFacultyByCourseIds() only queried 'editingteacher', missing faculty with the 'teacher' Moodle role. Now queries both roles via the EnrollmentRole enum, matching the pattern already used in FacultyService. Closes #202, Closes #203 https://claude.ai/code/session_01LNfjqKQstgZ7skvJ2oXAkf Co-authored-by: Claude <noreply@anthropic.com> * FAC-93 perf: limit fields in getFacultyByCourseIds query #207 Drop eager Course populate (FK already on Enrollment) and restrict User populate to the 5 columns actually used in the mapping, reducing DB I/O. https://claude.ai/code/session_01U2k8WWJ9EJ9b3Rzg4Vq81S Co-authored-by: Claude <noreply@anthropic.com> * FAC-94 refactor: enrollment test relies on implicit Promise.all mock call order #208 Replace mockResolvedValueOnce chains with mockImplementation that matches on entity class (Enrollment vs QuestionnaireSubmission), making the test resilient to Promise.all reordering in the service. Closes #205 https://claude.ai/code/session_012UsMjq61dUvFhErpChg9cR Co-authored-by: Claude <noreply@anthropic.com> * FAC-95 FAC-96 refactor: curriculum service filter type safety and search deduplication#212 Replace `Record<string, unknown>` with typed `FilterQuery<Department>` and `FilterQuery<Program>` intermediates, removing all `as FilterQuery<T>` casts in ListPrograms and ListCourses (#210). Extract duplicated $and/$or/$ilike search block into a shared ApplySearchFilter helper used by all three list methods (#211). Closes #210 Closes #211 https://claude.ai/code/session_01Y6eWhKjmgLfPJQMJa4VQQB Co-authored-by: Claude <noreply@anthropic.com> * FAC-97 perf: add pagination to curriculum list endpoints#213 Curriculum list endpoints (/departments, /programs, /courses) returned unbounded result sets. Add page/limit query params via PaginationQueryDto and return paginated { data, meta } responses using findAndCount, matching the pattern used by admin, faculty, and dimensions modules. https://claude.ai/code/session_01Ug6JLCMPhy42Skcq7tWyVV Co-authored-by: Claude <noreply@anthropic.com> * FAC-98 feat: add audit trail MVP (#215) * docs: add audit trail MVP tech-spec WIP Initialize tech-spec for audit trail feature covering auth events, admin actions, and sensitive data mutations via BullMQ queue. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: update audit trail tech-spec with deep investigation findings Add detailed file references, code patterns, MVP endpoint targets, and testing strategy from Step 2 codebase investigation. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: generate full implementation plan for audit trail tech-spec Add 20 tasks across 5 phases, 11 acceptance criteria, testing strategy, and risk analysis. Step 3 of quick-spec workflow complete. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: finalize audit trail MVP tech-spec as ready-for-dev Rename WIP to tech-spec-audit-trail-mvp.md and mark status as ready-for-dev. Quick-spec workflow complete (steps 1-4). https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply pre-mortem hardening to audit trail tech-spec Add 6 preventions from pre-mortem analysis: @optional() AuditService injection in AuthService, Logger.warn on failed emissions, attempts:1 on audit jobs, CLS null-safety warnings, bulk ingestion documentation, and full payload logging on processor failures. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: add team refinements to audit trail tech-spec Log full emission params on failure, add @optional() undefined test case for auth service, add CLS null-metadata warning test for interceptor. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply all 23 adversarial review fixes to audit trail spec Critical: @global() AuditModule, explicit interceptor stacks on all MVP endpoints, fix DI scoping for host modules. High: move login/refresh to direct emit, extend @Audited() with resource option, add AuditAction enum, tap vs finalize docs. Medium: drop repo stub, fix occurredAt, add attempts:1 in code, separate browserName/os fields, fix role claim, complete file list. Low: clarify emit location, entity array registration, test Redis, interceptor ordering, endpoint count. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 2 adversarial review fixes to audit trail spec Fix 14 findings: emit audit AFTER transactions with void (fire-and- forget), replace existing @UseInterceptors instead of adding duplicates, AuditInterceptor falls back to JWT payload (no CurrentUserInterceptor required), import AppClsModule instead of CommonModule, PII-safe logging in failed handler, PostgreSQL DEFAULT on occurred_at, fix AC 4 syntax and Task 6 Zod contradiction, document empty metadata for /ingest and null resourceId for created-by-action scenarios. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 3 adversarial review fixes to audit trail spec Critical: add explicit transaction refactoring pattern with code snippets for Login() and RefreshToken() — emit after transaction returns, not inside. High: correct @optional() resilience claim, fix resourceType for wipe endpoint, add per-endpoint interceptor bullets, acknowledge logout MetaDataInterceptor gap. Medium: add metadata schema table, full UUID regex, AuthenticatedRequest type, justify @global() convention, full FileInterceptor config, fix /ingest CurrentUserInterceptor contradiction. Low: create()+flush() pattern, processor log counter, fix test description, base class name. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 4 adversarial review fixes to audit trail spec Fix metadata table accuracy (4 rows showed wrong shapes), add null- safe destructure for RequestMetadataService in direct emit, strategy Name fallback for minification, type action as AuditAction end-to-end, consistent PII redaction policy, add CurrentUserInterceptor to POST /submissions (highest-volume), make occurredAt required (no JS default), document MetaDataInterceptor DI resolution, clarify attempts config, note create-action resourceId gap, Redis failed job monitoring guidance, actorId is plain string not @manytoone, logging side effect note. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * FAC-98 feat: add audit trail MVP Implement append-only AuditLog entity backed by a BullMQ AUDIT queue with dual emission paths — interceptor for authenticated endpoints and direct emit for auth events where CLS context is unavailable. - AuditLog entity (no CustomBaseEntity, no soft delete, no FK to User) - AuditService with fire-and-forget Emit() that never breaks requests - @Audited() decorator + AuditInterceptor (RxJS tap, post-response) - @global() AuditModule — cross-cutting concern, no explicit imports - 11 MVP endpoints tagged: auth (login/logout/refresh), moodle sync, questionnaire (submit/ingest/wipe), analysis (create/confirm/cancel) - Sanitized failure metadata (fixed reason codes, no raw error.message) - Metadata size cap (4KB) on interceptor-captured params/query - actorId index for "what did user X do?" queries - Shared audit test helpers to DRY controller spec boilerplate - 622 tests passing, lint clean --------- Co-authored-by: Claude <noreply@anthropic.com> --------- Co-authored-by: Claude <noreply@anthropic.com>

@isnotempty

… Trail MVP (#217) * chore: add /audit skill with auto GitHub issue creation * FAC-87 FAC-88 fix: async bcrypt and expired token filtering in RefreshToken #199 Replace bcrypt.compareSync with concurrent async bcrypt.compare via Promise.all to avoid blocking the event loop (#195). Add expiresAt filter to the DB query so expired tokens are excluded before running expensive bcrypt comparisons (#197). Closes #195 Closes #197 https://claude.ai/code/session_01XttX7NLr6HY7TDLddDk2AG Co-authored-by: Claude <noreply@anthropic.com> * FAC-89 fix: username logged in plaintext in Moodle login error path #200 Closes #198 * FAC-90 fix: LoginRequest DTO missing @isnotempty and length constraints #201 Empty strings and arbitrarily long payloads were passing validation on the login endpoint. Add @isnotempty() and @maxlength() decorators to both username and password fields, consistent with RefreshTokenRequestBody. Add DTO validation tests covering empty and oversized inputs. Closes #196 * FAC-91 FAC-92 fix: include teacher role in faculty lookup and use EnrollmentRole enum #206 getFacultyByCourseIds() only queried 'editingteacher', missing faculty with the 'teacher' Moodle role. Now queries both roles via the EnrollmentRole enum, matching the pattern already used in FacultyService. Closes #202, Closes #203 https://claude.ai/code/session_01LNfjqKQstgZ7skvJ2oXAkf Co-authored-by: Claude <noreply@anthropic.com> * FAC-93 perf: limit fields in getFacultyByCourseIds query #207 Drop eager Course populate (FK already on Enrollment) and restrict User populate to the 5 columns actually used in the mapping, reducing DB I/O. https://claude.ai/code/session_01U2k8WWJ9EJ9b3Rzg4Vq81S Co-authored-by: Claude <noreply@anthropic.com> * FAC-94 refactor: enrollment test relies on implicit Promise.all mock call order #208 Replace mockResolvedValueOnce chains with mockImplementation that matches on entity class (Enrollment vs QuestionnaireSubmission), making the test resilient to Promise.all reordering in the service. Closes #205 https://claude.ai/code/session_012UsMjq61dUvFhErpChg9cR Co-authored-by: Claude <noreply@anthropic.com> * FAC-95 FAC-96 refactor: curriculum service filter type safety and search deduplication#212 Replace `Record<string, unknown>` with typed `FilterQuery<Department>` and `FilterQuery<Program>` intermediates, removing all `as FilterQuery<T>` casts in ListPrograms and ListCourses (#210). Extract duplicated $and/$or/$ilike search block into a shared ApplySearchFilter helper used by all three list methods (#211). Closes #210 Closes #211 https://claude.ai/code/session_01Y6eWhKjmgLfPJQMJa4VQQB Co-authored-by: Claude <noreply@anthropic.com> * FAC-97 perf: add pagination to curriculum list endpoints#213 Curriculum list endpoints (/departments, /programs, /courses) returned unbounded result sets. Add page/limit query params via PaginationQueryDto and return paginated { data, meta } responses using findAndCount, matching the pattern used by admin, faculty, and dimensions modules. https://claude.ai/code/session_01Ug6JLCMPhy42Skcq7tWyVV Co-authored-by: Claude <noreply@anthropic.com> * FAC-98 feat: add audit trail MVP (#215) * docs: add audit trail MVP tech-spec WIP Initialize tech-spec for audit trail feature covering auth events, admin actions, and sensitive data mutations via BullMQ queue. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: update audit trail tech-spec with deep investigation findings Add detailed file references, code patterns, MVP endpoint targets, and testing strategy from Step 2 codebase investigation. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: generate full implementation plan for audit trail tech-spec Add 20 tasks across 5 phases, 11 acceptance criteria, testing strategy, and risk analysis. Step 3 of quick-spec workflow complete. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: finalize audit trail MVP tech-spec as ready-for-dev Rename WIP to tech-spec-audit-trail-mvp.md and mark status as ready-for-dev. Quick-spec workflow complete (steps 1-4). https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply pre-mortem hardening to audit trail tech-spec Add 6 preventions from pre-mortem analysis: @optional() AuditService injection in AuthService, Logger.warn on failed emissions, attempts:1 on audit jobs, CLS null-safety warnings, bulk ingestion documentation, and full payload logging on processor failures. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: add team refinements to audit trail tech-spec Log full emission params on failure, add @optional() undefined test case for auth service, add CLS null-metadata warning test for interceptor. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply all 23 adversarial review fixes to audit trail spec Critical: @global() AuditModule, explicit interceptor stacks on all MVP endpoints, fix DI scoping for host modules. High: move login/refresh to direct emit, extend @Audited() with resource option, add AuditAction enum, tap vs finalize docs. Medium: drop repo stub, fix occurredAt, add attempts:1 in code, separate browserName/os fields, fix role claim, complete file list. Low: clarify emit location, entity array registration, test Redis, interceptor ordering, endpoint count. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 2 adversarial review fixes to audit trail spec Fix 14 findings: emit audit AFTER transactions with void (fire-and- forget), replace existing @UseInterceptors instead of adding duplicates, AuditInterceptor falls back to JWT payload (no CurrentUserInterceptor required), import AppClsModule instead of CommonModule, PII-safe logging in failed handler, PostgreSQL DEFAULT on occurred_at, fix AC 4 syntax and Task 6 Zod contradiction, document empty metadata for /ingest and null resourceId for created-by-action scenarios. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 3 adversarial review fixes to audit trail spec Critical: add explicit transaction refactoring pattern with code snippets for Login() and RefreshToken() — emit after transaction returns, not inside. High: correct @optional() resilience claim, fix resourceType for wipe endpoint, add per-endpoint interceptor bullets, acknowledge logout MetaDataInterceptor gap. Medium: add metadata schema table, full UUID regex, AuthenticatedRequest type, justify @global() convention, full FileInterceptor config, fix /ingest CurrentUserInterceptor contradiction. Low: create()+flush() pattern, processor log counter, fix test description, base class name. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 4 adversarial review fixes to audit trail spec Fix metadata table accuracy (4 rows showed wrong shapes), add null- safe destructure for RequestMetadataService in direct emit, strategy Name fallback for minification, type action as AuditAction end-to-end, consistent PII redaction policy, add CurrentUserInterceptor to POST /submissions (highest-volume), make occurredAt required (no JS default), document MetaDataInterceptor DI resolution, clarify attempts config, note create-action resourceId gap, Redis failed job monitoring guidance, actorId is plain string not @manytoone, logging side effect note. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * FAC-98 feat: add audit trail MVP Implement append-only AuditLog entity backed by a BullMQ AUDIT queue with dual emission paths — interceptor for authenticated endpoints and direct emit for auth events where CLS context is unavailable. - AuditLog entity (no CustomBaseEntity, no soft delete, no FK to User) - AuditService with fire-and-forget Emit() that never breaks requests - @Audited() decorator + AuditInterceptor (RxJS tap, post-response) - @global() AuditModule — cross-cutting concern, no explicit imports - 11 MVP endpoints tagged: auth (login/logout/refresh), moodle sync, questionnaire (submit/ingest/wipe), analysis (create/confirm/cancel) - Sanitized failure metadata (fixed reason codes, no raw error.message) - Metadata size cap (4KB) on interceptor-captured params/query - actorId index for "what did user X do?" queries - Shared audit test helpers to DRY controller spec boilerplate - 622 tests passing, lint clean --------- Co-authored-by: Claude <noreply@anthropic.com> --------- Co-authored-by: Claude <noreply@anthropic.com>

@isnotempty

…agination & Audit Trail MVP (#218) * chore: add /audit skill with auto GitHub issue creation * FAC-87 FAC-88 fix: async bcrypt and expired token filtering in RefreshToken #199 Replace bcrypt.compareSync with concurrent async bcrypt.compare via Promise.all to avoid blocking the event loop (#195). Add expiresAt filter to the DB query so expired tokens are excluded before running expensive bcrypt comparisons (#197). Closes #195 Closes #197 https://claude.ai/code/session_01XttX7NLr6HY7TDLddDk2AG * FAC-89 fix: username logged in plaintext in Moodle login error path #200 Closes #198 * FAC-90 fix: LoginRequest DTO missing @isnotempty and length constraints #201 Empty strings and arbitrarily long payloads were passing validation on the login endpoint. Add @isnotempty() and @maxlength() decorators to both username and password fields, consistent with RefreshTokenRequestBody. Add DTO validation tests covering empty and oversized inputs. Closes #196 * FAC-91 FAC-92 fix: include teacher role in faculty lookup and use EnrollmentRole enum #206 getFacultyByCourseIds() only queried 'editingteacher', missing faculty with the 'teacher' Moodle role. Now queries both roles via the EnrollmentRole enum, matching the pattern already used in FacultyService. Closes #202, Closes #203 https://claude.ai/code/session_01LNfjqKQstgZ7skvJ2oXAkf * FAC-93 perf: limit fields in getFacultyByCourseIds query #207 Drop eager Course populate (FK already on Enrollment) and restrict User populate to the 5 columns actually used in the mapping, reducing DB I/O. https://claude.ai/code/session_01U2k8WWJ9EJ9b3Rzg4Vq81S * FAC-94 refactor: enrollment test relies on implicit Promise.all mock call order #208 Replace mockResolvedValueOnce chains with mockImplementation that matches on entity class (Enrollment vs QuestionnaireSubmission), making the test resilient to Promise.all reordering in the service. Closes #205 https://claude.ai/code/session_012UsMjq61dUvFhErpChg9cR * FAC-95 FAC-96 refactor: curriculum service filter type safety and search deduplication#212 Replace `Record<string, unknown>` with typed `FilterQuery<Department>` and `FilterQuery<Program>` intermediates, removing all `as FilterQuery<T>` casts in ListPrograms and ListCourses (#210). Extract duplicated $and/$or/$ilike search block into a shared ApplySearchFilter helper used by all three list methods (#211). Closes #210 Closes #211 https://claude.ai/code/session_01Y6eWhKjmgLfPJQMJa4VQQB * FAC-97 perf: add pagination to curriculum list endpoints#213 Curriculum list endpoints (/departments, /programs, /courses) returned unbounded result sets. Add page/limit query params via PaginationQueryDto and return paginated { data, meta } responses using findAndCount, matching the pattern used by admin, faculty, and dimensions modules. https://claude.ai/code/session_01Ug6JLCMPhy42Skcq7tWyVV * FAC-98 feat: add audit trail MVP (#215) * docs: add audit trail MVP tech-spec WIP Initialize tech-spec for audit trail feature covering auth events, admin actions, and sensitive data mutations via BullMQ queue. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: update audit trail tech-spec with deep investigation findings Add detailed file references, code patterns, MVP endpoint targets, and testing strategy from Step 2 codebase investigation. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: generate full implementation plan for audit trail tech-spec Add 20 tasks across 5 phases, 11 acceptance criteria, testing strategy, and risk analysis. Step 3 of quick-spec workflow complete. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: finalize audit trail MVP tech-spec as ready-for-dev Rename WIP to tech-spec-audit-trail-mvp.md and mark status as ready-for-dev. Quick-spec workflow complete (steps 1-4). https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply pre-mortem hardening to audit trail tech-spec Add 6 preventions from pre-mortem analysis: @optional() AuditService injection in AuthService, Logger.warn on failed emissions, attempts:1 on audit jobs, CLS null-safety warnings, bulk ingestion documentation, and full payload logging on processor failures. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: add team refinements to audit trail tech-spec Log full emission params on failure, add @optional() undefined test case for auth service, add CLS null-metadata warning test for interceptor. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply all 23 adversarial review fixes to audit trail spec Critical: @global() AuditModule, explicit interceptor stacks on all MVP endpoints, fix DI scoping for host modules. High: move login/refresh to direct emit, extend @Audited() with resource option, add AuditAction enum, tap vs finalize docs. Medium: drop repo stub, fix occurredAt, add attempts:1 in code, separate browserName/os fields, fix role claim, complete file list. Low: clarify emit location, entity array registration, test Redis, interceptor ordering, endpoint count. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 2 adversarial review fixes to audit trail spec Fix 14 findings: emit audit AFTER transactions with void (fire-and- forget), replace existing @UseInterceptors instead of adding duplicates, AuditInterceptor falls back to JWT payload (no CurrentUserInterceptor required), import AppClsModule instead of CommonModule, PII-safe logging in failed handler, PostgreSQL DEFAULT on occurred_at, fix AC 4 syntax and Task 6 Zod contradiction, document empty metadata for /ingest and null resourceId for created-by-action scenarios. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 3 adversarial review fixes to audit trail spec Critical: add explicit transaction refactoring pattern with code snippets for Login() and RefreshToken() — emit after transaction returns, not inside. High: correct @optional() resilience claim, fix resourceType for wipe endpoint, add per-endpoint interceptor bullets, acknowledge logout MetaDataInterceptor gap. Medium: add metadata schema table, full UUID regex, AuthenticatedRequest type, justify @global() convention, full FileInterceptor config, fix /ingest CurrentUserInterceptor contradiction. Low: create()+flush() pattern, processor log counter, fix test description, base class name. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * docs: apply round 4 adversarial review fixes to audit trail spec Fix metadata table accuracy (4 rows showed wrong shapes), add null- safe destructure for RequestMetadataService in direct emit, strategy Name fallback for minification, type action as AuditAction end-to-end, consistent PII redaction policy, add CurrentUserInterceptor to POST /submissions (highest-volume), make occurredAt required (no JS default), document MetaDataInterceptor DI resolution, clarify attempts config, note create-action resourceId gap, Redis failed job monitoring guidance, actorId is plain string not @manytoone, logging side effect note. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP * FAC-98 feat: add audit trail MVP Implement append-only AuditLog entity backed by a BullMQ AUDIT queue with dual emission paths — interceptor for authenticated endpoints and direct emit for auth events where CLS context is unavailable. - AuditLog entity (no CustomBaseEntity, no soft delete, no FK to User) - AuditService with fire-and-forget Emit() that never breaks requests - @Audited() decorator + AuditInterceptor (RxJS tap, post-response) - @global() AuditModule — cross-cutting concern, no explicit imports - 11 MVP endpoints tagged: auth (login/logout/refresh), moodle sync, questionnaire (submit/ingest/wipe), analysis (create/confirm/cancel) - Sanitized failure metadata (fixed reason codes, no raw error.message) - Metadata size cap (4KB) on interceptor-captured params/query - actorId index for "what did user X do?" queries - Shared audit test helpers to DRY controller spec boilerplate - 622 tests passing, lint clean --------- --------- Co-authored-by: Claude <noreply@anthropic.com>

claude and others added 11 commits March 29, 2026 18:18

docs: add audit trail MVP tech-spec WIP

e7497c4

Initialize tech-spec for audit trail feature covering auth events, admin actions, and sensitive data mutations via BullMQ queue. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

docs: update audit trail tech-spec with deep investigation findings

e27028a

Add detailed file references, code patterns, MVP endpoint targets, and testing strategy from Step 2 codebase investigation. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

docs: generate full implementation plan for audit trail tech-spec

592f216

Add 20 tasks across 5 phases, 11 acceptance criteria, testing strategy, and risk analysis. Step 3 of quick-spec workflow complete. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

docs: finalize audit trail MVP tech-spec as ready-for-dev

49f67a7

Rename WIP to tech-spec-audit-trail-mvp.md and mark status as ready-for-dev. Quick-spec workflow complete (steps 1-4). https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

docs: add team refinements to audit trail tech-spec

88db8ef

Log full emission params on failure, add @optional() undefined test case for auth service, add CLS null-metadata warning test for interceptor. https://claude.ai/code/session_01EXwPk6t54GxvuoMNtD1SSP

y4nder linked an issue Mar 29, 2026 that may be closed by this pull request

FAC-98 feat: add audit trail MVP #214

Closed

y4nder self-assigned this Mar 29, 2026

y4nder merged commit 5c3a9a4 into develop Mar 29, 2026
2 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

FAC-98 feat: add audit trail MVP#215

FAC-98 feat: add audit trail MVP#215
y4nder merged 11 commits intodevelopfrom
FAC-98/feat/add-audit-trail-mvp

y4nder commented Mar 29, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

y4nder commented Mar 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

New Files (14)

Modified Files (18)

Key Design Decisions

Test plan

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

y4nder commented Mar 29, 2026 •

edited

Loading