Problem Statement
The default tasks-template.md lists tasks sequentially with no dependency metadata. Agents cannot determine which tasks are independent and safe to run concurrently.
Proposed Solution
Add three constructs to the tasks template:
1. (depends on ...) suffix on each task
- [ ] T001 Create project structure
- [ ] T002 Initialize project with dependencies (depends on T001)
- [ ] T003 Configure linting tools (depends on T001)
Tasks with no dependencies omit the suffix. This is required for any task that depends on another.
2. Dependency DAG section at the bottom of tasks.md
## Dependencies & Execution Order
Wave 1: T001 (no deps)
Wave 2: T002, T003 (T001 done → parallel)
Wave 3: T004, T005, T006 (T002 done → parallel agents)
...
Alternatives Considered
No response
Component
Spec templates (BDD, Testing Strategy, etc.)
AI Agent (if applicable)
None
Use Cases
No response
Acceptance Criteria
No response
Additional Context
Why
/implement can schedule tasks optimally instead of strict sequential execution
/taskstoissues can generate GitHub Issues with "blocked by" references
Problem Statement
The default
tasks-template.mdlists tasks sequentially with no dependency metadata. Agents cannot determine which tasks are independent and safe to run concurrently.Proposed Solution
Add three constructs to the tasks template:
1.
(depends on ...)suffix on each taskTasks with no dependencies omit the suffix. This is required for any task that depends on another.
2. Dependency DAG section at the bottom of tasks.md
Alternatives Considered
No response
Component
Spec templates (BDD, Testing Strategy, etc.)
AI Agent (if applicable)
None
Use Cases
No response
Acceptance Criteria
No response
Additional Context
Why
/implementcan schedule tasks optimally instead of strict sequential execution/taskstoissuescan generate GitHub Issues with "blocked by" references