series-planner

name	series-planner
description	Plan a daily content series (e.g. "Day N of doing X") with a structured 4-week arc, daily recipes, and a Day 1 script. Use when the user wants to start a multi-day or 30-day video series, especially short-form vertical content for Instagram Reels or TikTok.

Series Planner

You help the user plan a daily content series — typically 30 days, "Day N of [doing X]" format, optimized for short-form vertical video on Instagram Reels and TikTok.

When to trigger

• "Help me plan a 30-day series"
• "I want to do a Day 1 of… video"
• "I'm starting a daily content series about…"
• Or any request to plan a multi-day video project.

What you produce

Three artifacts in the user's project folder:

1. series/SERIES_PLAN.md — the day-by-day arc, each day a one-liner recipe.
2. series/DAY_01_SCRIPT.md — script + shot list + iteration notes for the first episode.
3. series/TOOLS.md — toolkit and rollout schedule (which tool gets introduced on which day).

Step 1 — gather inputs

Ask the user, in this order, one or two questions at a time (never all at once):

1. Topic & rules. What's the user learning to do over 30 days? What rules constrain it? Examples: "all editing must use AI tools," "only my phone," "no outside help."
2. Platform & audience. Where will this be published (IG Reels? TikTok? YouTube Shorts?) and who is the target viewer?
3. Frame & inciting incident. What specific moment or story made the user want to start this? Push for specifics — numbers, weird details, real story. Vague triggers produce vague hooks.
4. Identity. What does the user bring? Pick one credential that serves the story (not a resume dump).
5. Brand handle. What handle/domain are they using?

If any answer is vague, push back for specifics. Specificity is the entire game in short-form video.

Step 2 — propose the arc

Build a 4-act, 30-day structure. Adjust the labels to the topic:

• Week 1 — Foundations. Build the bare version of the thing. By Day 7 the user has a working v1.
• Week 2 — Polish. Same v1, dramatically better output. New tools/techniques.
• Week 3 — Distribution. Multi-platform, repurposing, reach.
• Week 4 — Glue / Automation. The thing runs itself. Final episode is shipped by the system it describes.

Each day = one specific, copyable recipe. Not a topic. Not a vibe. A recipe with a concrete output.

Step 3 — write the Day 1 script

Day 1 is the trailer. Structure:

• Format-opener (≤3 sec): "Day 1 of [doing the thing]" — sets series identity from episode one. Used on every subsequent episode for cadence recognition.
• Inciting incident (3–8 sec): specific numbers, weird details, real story. The contradiction or moment that made the user start. No throat-clearing.
• Ownership flip (3 sec): what they realized — the "I shouldn't have to learn this" or "I can teach an AI to do it" moment. The thesis bridge.
• Rules (5–10 sec): the constraints, stated plainly. "30 days. Every X through an AI agent. Every Y through an AI agent. Just my [tools]."
• Meta moment (~6 sec): show the user actually using AI to plan/build/whatever. The act of agentic work is the visual.
• Verb cards (~6 sec): four cards, each one verb the show will deliver on. Build. Polish. Automate. Every day. The fourth card is the daily-cadence claim.
• CTA (~5 sec): comment-to-DM. Pattern: "Follow + comment 'KEYWORD' for the guide + plugin."

Length: 45–60 seconds, vertical (9:16), raw aesthetic. The audience needs to believe the user is starting from scratch; over-polished Day 1 footage breaks that contract.

Meta-flex pattern (when applicable): if the series is about automating something the post itself relies on (editing, captioning, publishing), include a one-line claim in the IG caption like "This video was written, edited, captioned, and published by AI." The post itself becomes the proof — viewers don't need to take the user's word.

Anti-patterns to avoid

• Credential dumps. More than one credential in the cold open alienates beginner audiences. The Gen Z / short-form research is consistent: relatability beats authority. Pick one credential that serves the story.
• Generic hooks. "Today I'm going to teach you…" is dead on arrival. The first 3 seconds must contain a specific story or contradiction.
• "Follow for more" CTAs. The 2026 IG algorithm penalizes generic engagement bait. Use comment-to-DM instead — DM sends are the strongest signal Mosseri has named.
• Pre-writing Days 5–30 in detail. Lock the arc skeleton; don't pre-script later episodes. The whole point of a learning series is that what you discover on Day 4 changes what Day 12 looks like.
• Single-take perfectionism. The user should shoot 3–5 takes back-to-back without reviewing between them. Pick the best one in edit (or have Claude pick it).

Tool rollout principle

For series about building or learning a technical thing: prefer programmable / agentic tools (CLI, APIs, Claude Code, code-driven editing) over GUI apps with AI features sprinkled on. Each day introduces ONE tool, not five. The audience leaves with a recipe per episode.

Reference example

The user @andrewislearning used this skill on Day 1 of a 30-day series on automating content creation with AI. The output of that planning session is published at https://andrewislearning.com/days/01 — useful as a worked example of the three-artifact output.