Create structured implementation plans with task breakdowns, change proposals, and verification steps. Use when starting new features, fixing bugs, refactoring code, or when user mentions "implementation plan", "design document", "technical spec", or needs to break down complex work into actionable tasks.
Systematically research codebases, design solutions, and create actionable implementation plans before writing code. This skill teaches AI agents to separate planning from execution, ensuring thorough context gathering and clear task definition.
The implementation planning workflow follows a two-phase approach:
Focus: Understanding the problem and designing the solution
Focus: Executing the approved plan
Planning Mode triggers:
Implementation Mode triggers:
Before writing any plan, gather comprehensive context using read-only operations:
Start with high-level overview:
.github/instructions/ related to the taskProgressive refinement:
Stopping criteria:
Tools for research:
read_file - Read project documentation and source filesfile_search - Find relevant files by pattern or contentgrep - Search for specific code patternslist_directory - Understand project structureCreate a structured markdown document that serves as the contract between planning and implementation phases.
File naming:
.github/implementations/implementation.instructions.md
.github/implementations/[module_name]_implementation.instructions.md
File location:
Store all implementation plans in .github/implementations/ directory for consistency.
YAML Frontmatter Structure:
---