creating-development-plans_skill

Overview

This skill creates structured development plans with phased task breakdowns, explicit requirements, and QA checklists. It produces an actionable DEVELOPMENT_PLAN.md that another senior agent or developer can follow to implement features. The output emphasises clarity, review checkpoints, and minimal over‑engineering.

How this skill works

I begin by gathering contextual information about the codebase, documentation and technology stack, then extract hard requirements and assumptions. The plan is organised into discrete phases with nested tasks, dependencies, self‑review gates and mandatory human review stops. A concise QA checklist and testing strategy are generated to ensure deliverables meet the stated requirements.

When to use it

• When you explicitly ask to create a development plan or document development requirements
• Before starting any non-trivial feature or integration work
• When onboarding a new developer or AI agent to an existing codebase
• When you need clear review checkpoints and quality gates defined
• When architectural trade-offs and integration points must be documented

Best practices

• Gather and list existing docs, tests and architecture notes before writing tasks
• Keep each phase independently testable and finish with a STOP for human review
• Avoid over‑engineering: prefer simple, well‑tested solutions over complex abstractions
• State assumptions and unknowns explicitly so they can be clarified before implementation
• Include project‑specific QA items and technology‑specific checks (linters, vetting, security)

Example use cases

• Producing a DEVELOPMENT_PLAN.md for a new feature in a Python microservice
• Defining phased work and QA for integrating an external API into an existing product
• Creating an onboarding plan so another AI agent can implement a module reliably
• Preparing a rollout plan for refactoring a legacy component with clear test gates
• Drafting testing and debugging protocols for a high‑risk production change