Overview

Agent Dispatch builds a dependency-aware, batched queue to run features concurrently without violating inter-feature relationships. It constructs a DAG from feature dependencies, sorts execution topologically, and batches work within a maxParallel constraint while assigning the appropriate agent types. The result is a batched dispatch queue ready for parallel execution by the automaker-feature-planner workflow.

How This Skill Works

Inputs include projectName, orderedPlans, maxParallel, and dependencyGraph. The process analyzes feature dependencies, builds a DAG, performs a topological sort to determine execution order, batches features into groups that respect concurrency limits and interdependencies, and assigns agent types to each feature. The outputs are a batched dispatch queue with agent assignments, intended for Stage 5 processing by automaker-feature-pipeline.js and Phase 3 batching by automaker-orchestrator.js.

When to Use It

• When multiple feature plans have interdependencies and you must respect dependencies.
• When you want to maximize parallel execution up to a maxParallel limit while preventing conflicts.
• When you need to batch features so no batch contains interdependent features.
• When you need explicit agent-type assignment per feature (e.g., Feature Planner).
• When preparing an executable queue for automaker-feature-pipeline.js Stage 5 and automaker-orchestrator.js Phase 3.

Quick Start

1. Step 1: Provide projectName, orderedPlans, maxParallel, and dependencyGraph to initialize the burst.
2. Step 2: Run the DAG construction, perform a topological sort, and create batches that respect maxParallel and dependencies.
3. Step 3: Assign agent types to each feature, output the batched dispatch queue, and feed into automaker-feature-pipeline.js (Stage 5) and automaker-orchestrator.js (Phase 3).

Best Practices

• Model all feature relationships in a dependency graph before batching to ensure accuracy.
• Set a realistic maxParallel based on system capacity and agent capabilities.
• Validate the DAG to ensure there are no cycles; re-validate after updates.
• Log batch composition and agent assignments for traceability and auditing.
• Test end-to-end with representative plans to verify topological order and batching integrity.

Example Use Cases

• Batching feature plans in a modular product where Feature A precedes Feature B due to data dependency.
• Dispatching parallel tasks for independent features while ensuring no cross-batch interdependencies.
• Assigning Feature Planner-like agents to features based on plan type and readiness.
• Optimizing throughput by batching up to maxParallel without violating dependencies in a CI-like workflow.
• Using DAG-based batching to prevent inter-feature collisions in a multi-agent automation environment.

Frequently Asked Questions