umbraco-upgrade-tasker

Umbraco Upgrade Tasker (15 → 17)

Convert an architecture plan into discrete, assignable, dev-ready tasks that a team can pick up and execute without needing to understand the full upgrade context.

When to Use

• After the architect skill has produced a migration plan
• When a tech lead needs to populate a project board
• When work needs to be distributed across a team
• As the final step in the agent orchestration pipeline

Inputs

• Architecture plan JSON from umbraco-upgrade-architect (preferred)
• OR: architecture plan markdown
• OR: audit findings + manual plan description
• Team context: size, roles, sprint length (optional, for sprint planning)

Core Principles

Each task must be independently understandable by a developer who hasn't read the full plan. This means:

1. Self-contained context: The task description explains WHY this change is needed, not just WHAT to do
2. Specific file references: Where possible, list the exact files that need to change
3. Acceptance criteria: Clear, testable conditions for "done"
4. Migration recipe reference: Link to the specific playbook recipe so the dev has step-by-step guidance
5. Dependencies are explicit: If task B requires task A, say so clearly

Workflow

Step 1: Load the Plan

Read the architecture plan (JSON or markdown). Extract:

• All work items from each phase
• The dependency chain between phases
• Risk levels and effort estimates
• Workstream assignments

Step 2: Decompose Work Items into Tasks

Each work item from the plan may become one or more tasks. The decomposition rules:

1:1 mapping (work item = task) when:

• The work is small (XS or S effort)
• It touches a single area
• One developer can complete it independently

1:N decomposition (work item → multiple tasks) when:

• The work item spans multiple files/projects
• It mixes code changes with configuration changes
• It has distinct "implement" and "verify" steps
• Different team members would work on different parts

Task Decomposition Rules

For each finding category, use these decomposition patterns:

Recompile tasks:

• Single task: "Recompile [component] against U17 packages"
• Include: which projects, expected outcome, how to verify

Find-and-Replace tasks:

• Single task per pattern type (group similar changes)
• Include: regex pattern, file glob, before/after example
• AC: "grep returns zero results for old pattern"

Refactor tasks:

• One task per logical unit of change
• Include: specific recipe reference from migration playbook
• Include: relevant test coverage expectations
• AC: "code compiles + unit tests pass + manual verification of [specific behavior]"

Rebuild tasks:

• Break into: research/spike → implement → integrate → test
• Each sub-task is independently assignable
• Flag rebuild tasks with "LARGE — may need refinement" if > L effort

Configure tasks:

• One task per configuration area
• Include: exact JSON/XML to add/change
• AC: "application starts successfully with new configuration"

Verify tasks:

• Pair with related code/config tasks
• Include: specific test scenarios to run
• AC: detailed test script or checklist

Step 3: Assign Workstreams

Every task gets a primary workstream:

Workstream	Roles	Typical Tasks
backend	.NET developers	C# changes, API updates, data access
backoffice	Frontend / Umbraco developers	Lit components, extension updates
devops	DevOps / infrastructure	CI/CD, Docker, deployment config
qa	QA / testers	Regression testing, content verification
admin	Tech lead / PM	License migration, decision-making

Step 4: Build Dependency Graph

Map task dependencies:

• Phase-level: Phase N tasks generally depend on Phase N-1
• Task-level: Some tasks within a phase have internal dependencies
• Parallel opportunities: Identify tasks that can run in parallel across workstreams

Express as:

TASK-001 → TASK-005 (TASK-005 requires TASK-001 complete)
TASK-002 ∥ TASK-003 (can run in parallel)

Step 5: Generate Output

Produce tasks in three formats:

5a. Markdown Task List

Use the task template at assets/task-template.md (local to this skill directory). For each task:

### TASK-XXX: [Title]

**Phase:** [N] — [Phase Name]
**Workstream:** [backend | backoffice | devops | qa | admin]
**Effort:** [XS | S | M | L | XL]
**Risk:** [LOW | MEDIUM | HIGH]
**Depends On:** [TASK-YYY, TASK-ZZZ] or None
**Source Findings:** [CS-001, CS-005]
**Playbook Reference:** migration-playbook.md#[section]

#### Context
[WHY this change is needed — 2-3 sentences a dev can read to understand the purpose]

#### What to Do
[Specific instructions — files to change, patterns to search, code to update]

#### Files Affected
- `src/path/to/file.cs` (line ~42)
- `src/path/to/other-file.cs`

#### Acceptance Criteria
- [ ] [Specific, testable criterion]
- [ ] [Another criterion]
- [ ] Code compiles without errors
- [ ] Existing tests pass

#### PR Description Template

What: [One-line summary] Why: [Breaking change reference] How: [Approach taken] Testing: [How this was verified] Related: TASK-XXX

5b. CSV for Project Management Import

Generate a CSV compatible with Jira/Azure DevOps/Linear import:

ID,Title,Description,Type,Priority,Effort,Workstream,Phase,Dependencies,Acceptance Criteria,Labels
TASK-001,"Update target framework to .NET 10","...","Task","Critical","S","devops","Phase 1","","Solution builds on .NET 10","umbraco-upgrade,framework"

Use scripts/generate-tasks.py (local to this skill directory) for CSV generation:

python3 scripts/generate-tasks.py tasks.json --format csv --output tasks.csv
python3 scripts/generate-tasks.py tasks.json --format azdo --output tasks-azdo.csv
python3 scripts/generate-tasks.py tasks.json --format jira --output tasks-jira.csv

5c. JSON (Structured)

{
  "task_metadata": {
    "project_name": "string",
    "total_tasks": 0,
    "total_effort_days": { "min": 0, "max": 0 },
    "generated_date": "ISO-8601"
  },
  "tasks": [
    {
      "id": "TASK-001",
      "title": "string",
      "phase": 1,
      "phase_name": "Framework & Infrastructure",
      "workstream": "devops",
      "effort": "S",
      "effort_hours": { "min": 1, "max": 4 },
      "risk": "LOW",
      "depends_on": [],
      "source_findings": ["PROJ-001"],
      "playbook_ref": "migration-playbook.md#1",
      "context": "string",
      "instructions": "string",
      "files_affected": ["src/MyProject/MyProject.csproj"],
      "acceptance_criteria": ["Solution builds targeting net10.0"],
      "labels": ["umbraco-upgrade", "framework"],
      "pr_template": "string"
    }
  ],
  "dependency_graph": {
    "TASK-001": { "depends_on": [], "blocks": ["TASK-005", "TASK-006"] },
    "TASK-005": { "depends_on": ["TASK-001"], "blocks": [] }
  },
  "workstream_summary": {
    "backend": { "task_count": 0, "total_effort_hours": { "min": 0, "max": 0 } },
    "backoffice": { "task_count": 0, "total_effort_hours": { "min": 0, "max": 0 } },
    "devops": { "task_count": 0, "total_effort_hours": { "min": 0, "max": 0 } },
    "qa": { "task_count": 0, "total_effort_hours": { "min": 0, "max": 0 } },
    "admin": { "task_count": 0, "total_effort_hours": { "min": 0, "max": 0 } }
  },
  "sprint_suggestions": [
    {
      "sprint": 1,
      "theme": "Foundation",
      "tasks": ["TASK-001", "TASK-002", "TASK-003"],
      "parallel_tracks": [
        { "workstream": "devops", "tasks": ["TASK-001"] },
        { "workstream": "admin", "tasks": ["TASK-002"] }
      ]
    }
  ]
}

Step 6: Save Outputs

Save to:

• upgrade-tasks/tasks.md — full markdown task list
• upgrade-tasks/tasks.csv — importable CSV
• upgrade-tasks/tasks.json — structured JSON
• upgrade-tasks/dependency-graph.md — visual dependency map

Agent Mode Instructions

When running as the tasker agent:

• Accept the architecture plan JSON from the architect agent
• Read the migration playbook at the plugin's references/migration-playbook.md (two directories up: ../../references/migration-playbook.md) for migration recipes to reference in tasks
• Generate all three output formats
• Include sprint suggestions if team context is provided
• Output files to the workspace directory

Effort → Hours Mapping

Size	Hours (min)	Hours (max)	Typical Work
XS	0.25	0.5	Config change, single find-replace
S	0.5	4	Small refactor, single file update
M	4	16	Multi-file refactor, API migration
L	16	40	Component rebuild, complex migration
XL	40	80+	Full rewrite (e.g., AngularJS → Lit)

Tips

• Number tasks sequentially within phases: TASK-101, TASK-102 (Phase 1), TASK-201, TASK-202 (Phase 2), etc.
• Always include a "Verify [area]" task after code change tasks — testing is work too
• Group find-and-replace tasks together when the same developer would do them in one sitting
• For XL tasks (rebuilds), add a "Spike: Research approach for [component]" task first
• CSV format should match the team's actual PM tool — ask about custom fields if needed
• Sprint suggestions assume 2-week sprints unless told otherwise