Task Quality Review
Folder: .github/skills/tsh-task-quality-reviewing/
Used by: Business Analyst
Performs a systematic quality analysis on an approved task list (epics and user stories) to identify gaps, missing edge cases, and improvement opportunities. Runs 10 domain-agnostic analysis passes and produces structured suggestions the user can individually accept or reject.
What It Produces
- Suggestions — Structured improvement proposals, each individually accept/reject.
- Domain model — An intermediate actor–entity–lifecycle map derived from the task list.
- Quality review report — Audit trail of all suggestions and decisions (
quality-review.md). - Updated task list — Accepted suggestions applied to
extracted-tasks.mdin-place.
Analysis Passes
The quality review runs 10 independent analysis passes:
| Pass | Category | What It Checks | Confidence |
|---|---|---|---|
| A | Entity Lifecycle Completeness | CRUD operations for every business entity | High |
| B | Cross-Feature State Validation | State checking when features consume shared entities | High |
| C | Bulk Operation Idempotency | Pre-existing data and partial failure handling | High |
| D | Actor Dashboard Completeness | Metrics, configuration, and history for management UIs | Medium |
| E | Precondition Guards | Feature unlock dependencies and prerequisite enforcement | High |
| F | Third-Party Boundary Clarity | External integration points and failure modes | Medium |
| G | Platform Operations Perspective | Admin/operator tooling and monitoring | Medium |
| H | Error State & Edge Case Coverage | Failure, empty, and boundary conditions | High |
| I | Notification & Communication Gaps | Missing notifications for cross-actor state changes | High |
| J | Domain-Specific Research | Industry patterns and compliance requirements | Low–Medium |
Suggestion Types
Each finding is classified into one of these action types:
| Action Type | Description |
|---|---|
ADD_ACCEPTANCE_CRITERION | Add a missing condition to an existing story |
MODIFY_STORY | Expand an existing story's scope |
ADD_TECHNICAL_NOTE | Add clarity or documentation to a story |
NEW_STORY | Create a new story for uncovered functionality |
NEW_EPIC | Create a new epic for a major capability gap |
Process
Step 1: Load Inputs
Collect the Gate 1-approved extracted-tasks.md, cleaned transcript, and any other source materials.
Step 2: Gather Jira Context (Optional)
If Atlassian tools are available, optionally fetch existing board context to cross-reference against.
Step 3: Build Domain Model
Construct a lightweight domain model from the task list: actors, entities (with lifecycle mapping), and relationships.
Step 4: Run Analysis Passes
Execute all 10 passes against the domain model and task list. Each pass produces zero or more findings.
Step 5: Classify Suggestions
Transform findings into structured suggestions with confidence levels, action types, and proposed changes.
Step 6: User Review (Gate 1.5)
Present suggestions one at a time for individual accept/reject decisions. Each suggestion is self-contained with full context.
Step 7: Apply Accepted Changes
Apply accepted suggestions to extracted-tasks.md and save the quality review report to quality-review.md.
Connected Skills
tsh-task-extracting— Provides the extracted tasks used as primary input.tsh-jira-task-formatting— Consumes the updated task list after quality review.