Can this skill work with my existing project management tools?

Yes, it is tool-agnostic and can adapt to native agent tools, GitHub Issues, file-based trackers like TASKS.md, or CLI-based task managers.

Why does the skill enforce a 'one active task' rule?

Limiting agents to one active task per execution context ensures high focus, reduces the risk of incomplete context, and provides an unambiguous audit trail of progress.

What makes an acceptance criterion 'binary'?

A criterion is binary if an outside observer can verify it as met or not met without subjective judgment, such as 'API returns a 200 status' rather than 'API responds quickly'.

What is the 'Slice-to-Task' decomposition method?

It is a methodology where each task represents a single testable behavior increment following a Red-Green-Commit cycle, ensuring continuous delivery of functional code.

Task Management for AI Agents

Name: Task Management for AI Agents
Author: jwilger

byjwilger

•

Productivity & Workflow

Decomposes complex software features into structured, trackable tasks with explicit dependencies and binary acceptance criteria.

This skill provides a robust framework for breaking down large software goals into manageable units, from high-level epics to granular, testable leaf tasks. By implementing a strict lifecycle and enforcing explicit dependency mapping, it ensures that AI agents maintain focus, provide measurable progress, and follow a 'Slice-to-Task' methodology that integrates seamlessly with TDD workflows. It is essential for teams looking to make AI progress visible, verifiable, and free of bottlenecks through clear Given/When/Then criteria and automated status tracking.

Key Features

012 GitHub stars

02Slice-to-Task methodology for incremental, test-driven delivery cycles

03Hierarchical work decomposition from epics down to granular leaf tasks

04Strict 'One Active Task' rule to maximize agent focus and reduce context switching

05Explicit dependency mapping to sequence work and identify unblocked tasks

06Binary Given/When/Then (GWT) acceptance criteria for objective verification

Use Cases

01Breaking down complex feature requests into a 'walking skeleton' and testable slices

02Standardizing work progress tracking across different tools like GitHub Issues or local task files

03Managing dependencies in multi-agent environments to prevent circular blocking

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jwilger/agent-skills task-management

For use in Claude.ai and ChatGPT

Download Skill