Can I modify the plan after it's generated?

Yes. If the generated plan isn't quite right, you can provide feedback. The skill will regenerate the entire atomic plan rather than executing a partial or incorrect sequence.

How does the parallelization batching work?

The skill identifies tasks with zero dependencies and places them in 'Batch 1'. Once those are cleared, tasks that were previously blocked form 'Batch 2', allowing you to see which parts of a project can run concurrently.

Does it handle circular dependencies?

Yes. Identifying circular dependencies is a core function of the skill. If a blocking relationship is circular, the orchestrator will surface it as a bug for you to address during the planning phase.

When should I use the 'ultrathink' mode?

You should trigger deep thinking for complex projects with more than 10 tasks, non-obvious cross-system dependencies, or when the cost of a sequencing failure is high.

Does Task Orchestrator execute code automatically?

No. The skill strictly follows a planning-only contract. It generates a numbered plan and requires explicit user confirmation (like 'yes' or 'proceed') before any subagents are invoked.

Task Orchestrator

Name: Task Orchestrator
Author: mahidalhan

bymahidalhan

•

生産性とワークフロー

Orchestrates complex multi-step tasks by mapping dependencies and grouping parallel workflows into optimized, human-approved execution plans.

Task Orchestrator is a high-level planning skill designed for managing sophisticated development projects that require precise sequencing. It analyzes complex requests to identify critical paths, detects circular dependencies, and groups non-blocking tasks into parallel batches to minimize execution latency. By enforcing a strict 'plan-then-confirm' protocol, it ensures that Claude never performs unauthorized actions, providing a transparent and atomic breakdown of work before any code is modified. It is particularly effective for large-scale refactors and feature implementations where 'deep thinking' modes are required to map out non-obvious system relationships.

主な機能

01Dependency Graph Visualization: Maps explicit task relationships to identify blockers.

02Parallel Execution Batching: Groups independent tasks to maximize efficiency.

0320 GitHub stars

04Deep Thinking Integration: Supports 'ultrathink' for complex 10+ task dependency analysis.

05Execution Gatekeeping: Prevents auto-execution by requiring explicit user confirmation.

06Atomic Task Decomposition: Breaks down large feature requests into granular, manageable steps.

ユースケース

01Refactoring multi-module systems where the order of operations is critical to system stability.

02Planning the simultaneous rollout of backend API changes and frontend UI updates.

03Identifying and organizing parallelization opportunities in complex build or deployment workflows.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mahidalhan/claude-hacks task-orchestrator

For use in Claude.ai and ChatGPT

Download Skill