All skills
task-dag-planner
ProfessionalRuntime top levelanalysis-agent: create a Task DAG from an accepted source-backed Brief for 2+ semantic tasks with dependency, parallel, integration, or ordering needs; skip single edits, accepted DAGs, and unanalysed requests.
- Group
- Intake, analysis and planning
- Routing mode
- evidence-only
- Supported roles
- MainAnalysisTaskReview
- Source
- SKILL.md
Use when
Trigger signals that make this skill the right owner
- explicit Task DAG request with accepted Engineering Brief
- accepted Brief has multiple tasks with a dependency parallel benefit integration need or release order
Do not use when
Anti-triggers — as binding as the triggers above
- single bounded edit
- task DAG already accepted
- unanalysed change request
Required inputs
What must be supplied before this skill can decide anything
- accepted Engineering Brief
- source evidence
- semantic and Primary Professional Skill boundaries
- resource and minimum sufficient Review Boundaries
Output contract
What this skill owes the next role
- Task Capsules with one Primary Professional Skill each
- dependency edges
- integration order
- minimum sufficient Review Boundaries preserving required Review Skills and Specialist obligations
Escalation signals
When this skill must hand the decision back
- scope, authority, or material risk exceeds the selected task contract
Related expertise
2 Layer 3 candidates may be loaded behind this skill. A single task loads at most three, and only when the current evidence justifies them.
Targeted references
Each reference carries its own load condition. None of them enter context by default.
| Reference | Load when | Do not load when | Required by |
|---|---|---|---|
| checklist.mddecision-checklist | A bounded review needs a quick DAG readiness and closure checklist | Detailed evidence map or executable node contract is required | Analysis |
| index.mdindex | competing task dag planner references require dependency, conflict, or output-fragment selection | the task dag planner root or a task-named reference already resolves selection | Analysis |
| planning-evidence-patterns.mdevidence-pattern | Closing graph validity, new-hypothesis, parallelization, rollback, or plan-execution consistency | Only task field shape is needed | Analysis |
| task-contract-patterns.mdbenchmark-pattern | Nodes must be executable by a fresh implementer or placeholder tasks need replacement | Existing task contracts already name exact scope, behavior, validation, and residual risk | Analysis |