AI Coding Planning Frameworks

Choose a framework by the artifact and workflow you need, not by a universal score. Some methods organize one implementation plan. Others maintain project state, feature specifications, issue graphs, or a full product-delivery sequence. The public same-case comparison repository applies these artifact models to one reliable-webhook-delivery brief. It preserves generated output for five runnable frameworks, a manual source-faithful Kiro example, tool versions, interruptions, and limits.

Framework	Primary unit	Main artifacts	Useful when
Superpowers	Design and implementation plan	Repository specs and code-level plans	You want a defined path from design review to detailed execution
GSD Core	Project, milestone, phase, and plan	A durable `.planning/` tree	Work spans sessions and needs explicit state, phases, and verification
GitHub Spec Kit	Feature	Constitution, spec, plan, tasks, supporting artifacts	You want an open feature directory with structured requirements and tasks
Kiro Specs	Feature or bug	Requirements or bug analysis, design, tasks	You want an IDE-integrated three-file specification workflow
Matt Pocock’s skills	Issue, question, spec, or ticket	Issue-tracker artifacts and focused skills	You want composable planning skills inside an existing engineering process
BMad Method	Product initiative through story	Brief, PRD, architecture, epics, stories	You need a broader role-oriented product and delivery method
Amazon Working Backwards PRFAQ	Proposed product or feature	Future press release, customer FAQ, internal FAQ	You want to test customer value before writing detailed requirements
ADR formats	Architectural decision	One Markdown decision record	You need to preserve a significant choice and its consequences

How to choose

Ask five questions:

Is the main object a feature, project phase, issue graph, delivery program, or decision?
Should artifacts live in the repository, an issue tracker, or an IDE-managed directory?
Does the team want a prescribed sequence or composable procedures?
Should tasks name exact code edits or remain outcome-focused?
What review and verification record must survive into the next session?

The profiles describe observed source behavior at a pinned commit or capture date. They do not claim that one framework produces better code. For direct decisions, see Superpowers vs GSD and GitHub Spec Kit vs Kiro Specs. Reviewed July 19, 2026. Researched and maintained by the Plannotator documentation team.

Templates

Frameworks

Comparisons

AI Coding Planning Frameworks

How to choose

​How to choose

How to choose