Autonomous Task Executor FAQs

Question 1

What happens if a task fails during execution?

Accepted Answer

It includes an automatic retry mechanism (default 3 attempts). If a task fails or only partially succeeds, the orchestrator provides the failure context to the next attempt to help the agent self-correct.

Question 2

How does this skill handle task dependencies?

Accepted Answer

It builds a dependency graph from your task list and organizes them into 'waves.' Each wave contains only those tasks whose dependencies have been successfully completed, ensuring a logical implementation order.

Question 3

Can I run multiple tasks at the same time?

Accepted Answer

Yes, the skill supports configurable parallelism (defaulting to 5 tasks) using background agents to execute independent tasks simultaneously within the same wave.

Question 4

What is Adaptive Verification?

Accepted Answer

The skill detects if a task was generated from a spec or is a general request. For spec tasks, it performs rigorous criterion-by-criterion checks; for general tasks, it uses inferred requirements to ensure the core change works.

Question 5

How do parallel agents stay consistent with each other?

Accepted Answer

All agents in an execution session share a central execution_context.md file where they record discovered patterns and decisions, which are then merged and shared with subsequent tasks.

Autonomous Task Executor

主要功能

使用场景

Autonomous Task Executor

主要功能

使用场景