Intelligent Task Decomposer FAQs

Question 1

Can I run multiple subtasks at the same time?

Accepted Answer

Yes. The skill generates a 'Parallel Execution Plan' that identifies independent tasks and provides an execution order for spawning subagents concurrently without risking merge conflicts.

Question 2

When should I use the decompose-task skill?

Accepted Answer

You should use it when a task approaches the 80K token threshold (roughly 40% of the context window), when a subagent is experiencing compaction events, or when a PLAN.md analysis reveals a scope larger than originally expected.

Question 3

What are considered 'good' split points for a task?

Accepted Answer

Effective split points occur between independent features, distinct architectural layers (like moving from model to controller), or between setup, implementation, and testing phases.

Question 4

Does decomposing a task cause me to lose my current work?

Accepted Answer

No. The skill includes a specific workflow to collect results from existing subagents and merge that partial progress into the newly created subtask branches.

Question 5

How does the skill handle task dependencies?

Accepted Answer

It generates a dependency graph that distinguishes between runtime dependencies and extraction dependencies, allowing tasks that write to different file sections to run in parallel while sequencing those that rely on prior outputs.

Intelligent Task Decomposer

Intelligent Task Decomposer

Key Features

Use Cases

Key Features

Use Cases