uc-audit-cross-check

name	uc-audit-cross-check
description	Batch audit all Use Cases, evaluate relevance to SRS, cross-check UC-SQL-JIRA gaps. Triggers: audit UC, cross-check, use case coverage, gap analysis, UC inventory, UC relevance, UC SQL check, missing UC, missing task, orphan column, UC completeness, SRS alignment, check UC, evaluate UC.
category	project-management
risk	safe
source	custom
date_added	2026-03-05

Skill: UC_AUDIT — Use Case Audit & Cross-Check

Purpose

Batch-read and audit all 26 Use Cases, evaluate each UC's relevance to the SRS, and cross-check the triad UC ↔ SQL Schema ↔ JIRA Tasks to find gaps, orphans, and inconsistencies. This is the only skill designed to work with the entire UC corpus at once.

When to Use

• Verify UC completeness: Are all needed UCs written? Are any redundant?
• Evaluate UC-SRS alignment: Does every UC trace back to an SRS functional requirement (HG-FUNC)?
• Cross-check UC ↔ SQL: Does every data field mentioned in UCs have a corresponding DB column? Are there orphan DB columns not covered by any UC?
• Cross-check UC ↔ JIRA: Does every UC have at least one JIRA Story? Are there JIRA Stories without a UC?
• After SRS changes, new UCs, or DB schema updates

---

Context Loading Protocol (MANDATORY)

[!IMPORTANT] Follow this strict 4-tier loading protocol. DO NOT skip tiers. DO NOT read full SRS.

Tier 1: Navigation (ALWAYS)

1. Read PM_REVIEW/MASTER_INDEX.md — Project GPS map
2. Read PM_REVIEW/Resources/SRS_INDEX.md — System-level context, HG-FUNC requirements

Tier 2: UC Corpus Overview

3. Read PM_REVIEW/Resources/UC/00_DANH_SACH_USE_CASE.md — Master UC list (26 UCs), platform mapping, relationships
4. Note UC count, module groupings, deleted UCs, and inter-UC relationships (include/extend/trigger chains)

Tier 3: Cross-Reference Sources

5. Read PM_REVIEW/Resources/TASK/JIRA/README.md — JIRA Index (16 Epics, 61 Stories, UC→Epic lookup)
6. Read PM_REVIEW/SQL SCRIPTS/README.md — DB Architecture overview (6 layers, ~20 tables)

Tier 4: Deep Dive (Progressive — per phase)

7. Read individual UC files only during Phase 1 analysis
8. Read specific SQL files only during Phase 3 cross-check
9. Read JIRA Sprint/Epic STORIES.md only when verifying task coverage

⛔ WHAT NOT TO DO

• ❌ DO NOT read the full SRS document — use SRS_INDEX
• ❌ DO NOT read the full JIRA CSV — use JIRA Index
• ❌ DO NOT read all SQL files at once — read per-layer as needed
• ❌ DO NOT read all UC files in a single batch — process module by module

---

Phase 1: UC Inventory & Quality Check

Goal: Build a structured inventory of ALL UCs with quality assessment.

Process

Step 1 — Read UC files module by module:

Read UCs in this order (8 modules, 26 files total):

1. Authentication/ (6 files: UC001, UC002, UC003, UC004, UC005, UC009)
2. Monitoring/ (3 files: UC006, UC007, UC008)
3. Emergency/ (4 files: UC010, UC011, UC014, UC015)
4. Analysis/ (2 files: UC016, UC017)
5. Sleep/ (2 files: UC020, UC021)
6. Admin/ (4 files: UC022, UC024, UC025, UC026)
7. Notification/ (2 files: UC030, UC031)
8. Device/ (3 files: UC040, UC041, UC042)

Step 2 — For each UC, extract and record:

Field	What to Extract
UC ID	From spec table header (e.g., UC001)
Name	UC title from spec table
Module	Parent folder name
Actors	Primary actors from spec table
Platform	Mobile / Admin Web / Both
Main Flow Steps	Count of steps in main flow table
Alt Flow Count	Count of alternative flow sections
Business Rules	List of BR codes (e.g., BR-001, BR-040-01)
Data Fields	Fields mentioned (email, password, device_id, etc.)
DB Tables Referenced	Tables explicitly mentioned in the UC text
NFR Categories	Performance / Security / Usability

Step 3 — Quality check per UC:

Read references/uc-analysis-checklist.md for detailed per-UC quality criteria.

Key checks:

• Does the UC have all required sections? (Spec table, Main Flow, Alt Flows, Business Rules, NFR)
• Is the actor an external entity (not "System")?
• Are main flow steps ≤ 10?
• Are alt flows properly numbered (referencing main flow step)?
• Are business rules specific (not generic copy-paste)?

---

Phase 2: SRS Relevance Assessment

Goal: Score each UC's alignment with SRS functional requirements.

Process

Step 1 — Build SRS→UC mapping:

1. Read SRS_INDEX.md → extract the Feature → Functional Requirement Mapping table (HG-FUNC-01 to HG-FUNC-11)
2. For each HG-FUNC, identify which UCs implement it
3. Background processes (HG-FUNC-01, 04, 10) correctly have no UC — mark as N/A

Step 2 — Score each UC's relevance:

Read references/uc-analysis-checklist.md → Relevance Scoring Rubric for detailed criteria.

Level	Criteria
CORE	UC directly implements a HG-FUNC requirement
SUPPORTING	UC enables core features (auth, config, profile)
MANAGEMENT	UC is for system admin/management
LOW	UC has weak connection to system goals

Step 3 — Flag issues:

• HG-FUNC without any covering UC → UNCOVERED_REQUIREMENT
• UC without any HG-FUNC connection and not SUPPORTING/MANAGEMENT → ORPHAN_UC

---

Phase 3: Cross-Check Matrix (UC ↔ SQL ↔ JIRA)

Goal: Find gaps and inconsistencies across the three pillars.

Process

Read references/cross-check-matrix.md for the detailed matrix template and gap classification.

Check A — UC ↔ JIRA:

1. Use JIRA Index UC → Epic Lookup table
2. For EVERY UC in the inventory, verify it appears in the lookup
3. For every JIRA Epic/Story, verify it has a corresponding UC (unless infra/setup tasks)

Flag types:

• UC_NO_TASK — UC exists but no JIRA Story covers it
• TASK_NO_UC — JIRA Story references a UC that doesn't exist
• EPIC_NO_UC — JIRA Epic has no UC mapping (acceptable for infra Epics: EP01, EP02, EP03, EP06)

Check B — UC ↔ SQL:

1. For each UC, identify the data fields mentioned in Main Flow + Business Rules
2. Map those fields to SQL table columns (using SQL README + actual SQL files)
3. Verify the column exists and the data type is compatible

Flag types:

• MISSING_COLUMN — UC mentions a field not present in any SQL table
• ORPHAN_COLUMN — SQL column not referenced (directly or indirectly) by any UC
• TYPE_MISMATCH — UC implies a data type different from the SQL column

Check C — Internal UC Consistency:

1. 00_DANH_SACH_USE_CASE.md vs actual files in UC/ folders — any mismatch?
2. UC/README.md stats vs 00_DANH_SACH_USE_CASE.md stats — any desync?
3. MASTER_INDEX UC Refs vs actual UC files — any missing references?
4. Inter-UC relationships (include/extend) — are referenced UCs present?

---

Output Protocol (MANDATORY)

• File: PM_REVIEW/UC_AUDIT_report.md (project-wide, not per Admin/Mobile)
• Template: Read references/report-template.md — MUST follow exactly
• Language: Vietnamese section headers
• Overwrite: If file exists → overwrite, add comparison section (see template)
• Sections: Summary → UC Inventory table → SRS Alignment → Cross-Check → Recommendations

---

Reference Documents

Name	Path	When to read
MASTER INDEX	PM_REVIEW/MASTER_INDEX.md	ALWAYS
SRS Index	PM_REVIEW/Resources/SRS_INDEX.md	ALWAYS
UC Master List	PM_REVIEW/Resources/UC/00_DANH_SACH_USE_CASE.md	ALWAYS
UC README	PM_REVIEW/Resources/UC/README.md	Phase 1
UC Files	PM_REVIEW/Resources/UC/{Module}/*.md	Phase 1 (module by module)
JIRA Index	PM_REVIEW/Resources/TASK/JIRA/README.md	ALWAYS
SQL README	PM_REVIEW/SQL SCRIPTS/README.md	Phase 3
SQL Files	PM_REVIEW/SQL SCRIPTS/0{N}_*.sql	Phase 3 (per-layer)
Analysis Checklist	references/uc-analysis-checklist.md	Phase 1
Cross-Check Matrix	references/cross-check-matrix.md	Phase 3
Report Template	references/report-template.md	Output phase

Integrated Skills (Bundled)

CRITICAL: Before executing any phase, read the relevant bundled skills from the skills/ subdirectory to inherit their analysis methodology:

Bundled Skill	Path	Use in Phase
Business Analyst	skills/business-analyst/SKILL.md	Phase 1 (structured analysis), Phase 3 (gap identification)
Product Manager Toolkit	skills/product-manager-toolkit/SKILL.md	Phase 2 (prioritization), Phase 3 (coverage analysis)

---

Rules

• Process UCs module by module — never dump all 26 at once
• Every UC must be read — do NOT skip any UC, even if it looks trivial
• Compare multiple sources — always triangulate (UC list vs files vs MASTER_INDEX vs JIRA)
• Flag ALL discrepancies — count mismatches, desync, missing items
• Report in Vietnamese — final report uses Vietnamese section headers
• Overwrite old report — if re-auditing, overwrite UC_AUDIT_report.md directly
• Update MASTER_INDEX — if UCs are found missing from module rows, flag for update