Skill: Cut Planner

Overview

In the cut-planner skill, the main thread reads clip summaries from a library to understand footage coverage, then works with the user to design a narrative plan markdown file for this specific cut.

The library's footage_summary, user_context, and individual summaries already capture the broad creative context — that was locked in during footage analysis (see "Start Footage Analysis" in AGENTS.md). If anything is confusing you can ask the user, but generally you can just use the summary files and library.yaml for context.

This skill runs in the main thread and does not use a sub-agent.

Asking the user

Whenever you need the user to pick from a discrete set of options — which library, whether to resume processing, target length, which concept, go/no-go on the plan — use the AskUserQuestion tool (or similar option-chip tool the host agent provides) instead of writing a bullet list in chat. Free-form conversation about the footage stays as plain chat; only the moments where you're offering a choice should use the tool.

When the options are libraries (or anything else with a natural "last touched" signal), order them by recency — most recently modified first. For libraries, look at the newest mtime among files inside each library directory (transcripts, summaries, plans, library.yaml itself). The library the user worked on yesterday should be the first chip, not the alphabetical winner.

Cut Planner Process

1. Verify all clips have visual transcripts and summaries