Section 2 · /create-tasks
From Spec to Actionable Tasks
The /create-tasks command transforms a clarified specification into an ordered, traceable tasks.md file ready for implementation.
Prerequisite
Clarification Gate passed — zero open items in requirements.md. User stories and acceptance criteria written.
Command Workflow
Validate the spec
Clarification Gate: open items = 0. Stop if not.
Identify user stories
Extract stories and acceptance criteria from requirements.md.
Atomic decomposition
Split into smallest verifiable units per story.
Order by dependency
Foundation → core logic → edge cases → verification.
Link to requirements
Every task gets a REQ-XXX reference.
Add verification steps
Checkbox criteria per task — objective done definition.
Write tasks.md
Commit to repo — version-controlled alongside requirements.md.
Atomic Decomposition Checklist
- Task title describes one outcome, not a whole feature
- Task can be verified without completing other tasks (or lists dependency)
- Task links to exactly one primary requirement
- Verification steps are checkboxes, not prose
- Task is small enough for one PR or one AI session
Risk Planning Matrix
During the Plan stage, identify what can go wrong, how to detect it, and how to mitigate — before tasks become code.
| What can go wrong? | How do we detect it? | Mitigation Strategy |
|---|---|---|
| Database failure | Metrics / Alerts | Rollback migration |
| Latency spike | Monitoring signals | Disable via flag |
| API misuse | Logs | Rate limit |
Include a risk matrix in your plan alongside tasks — especially for features touching databases, APIs, or external services.
Anti-Patterns
| Avoid | Instead |
|---|---|
| "Build messaging feature" | "Design message data structure" |
| Tasks with no requirement link | Every task → REQ-XXX |
| No verification steps | Checkbox criteria per task |
| Flat unordered list | Groups by dependency |
| tasks.md in chat only | Commit tasks.md to repo |
| /create-tasks before gate passes | Resolve open items first |