Planning

This is the orchestrator for model customization workflows in Claude Code. It asks clarifying questions to understand what you're trying to do, then generates a structured plan that chains together the right skills in the right order. The key insight is the workflow choice gate: it won't let you skip past deciding between evaluate-first and direct fine-tuning, because most people don't know evaluate-first is an option. Once you approve a plan, it executes each task by loading the relevant skill and updating a PLAN.md file as it goes. It's opinionated about keeping plans short and surfacing constraints early, like flagging region limitations when you pick a Nova model. Always activates alongside other customization skills, even for narrow asks like reviewing data formats.

skills.sh installs: 168
GitHub stars: 938
last updated: Jun 11, 2026

View on GitHub

Install

$ npx -y skills add awslabs/agent-plugins --skill planning --agent claude-code

Installs into .claude/skills of the current project.

Files

Principles

One question at a time. Each question should resolve a branching decision in the plan. Avoid generic or out-of-domain questions.
Surface constraints early. If a user decision would constrain downstream options, flag it before the plan is finalized.
Keep plans short. Only include tasks that are necessary for the user's stated goal.
Don't ask what you already know. Check conversation history and project files before asking the user.