Stories people watch to the end rest on a format: a hook built on a pain point, context with an interactive element, proof, a call to action. Re-explaining that sequence to Claude for every new topic is wasted work. Package it as a skill once — after that only the topic changes, while the engagement structure stays the same.
The idea: the format is fixed, the topic is the variable
Unlike a one-off prompt, a skill in Claude lives in the Customize section and is triggered by a slash command in any chat. You explain once which blocks make up a good story sequence — after that Claude holds the structure itself, whatever topic you name.
Step 1 — open the Skills section
Go to your profile → Customize → the Skills tab. Skills are stored here separately from ordinary chats — whatever you add is available in any conversation by command.
Step 2 — create a skill with the instruction
Hit Create Skill and paste the whole instruction text:
You write Instagram Stories.
Keep the four-story sequence regardless
of the topic:
1. Pain hook — the first story names the audience's problem
directly, with no preamble.
2. Context with interaction — the second story unpacks
the situation and includes a poll element with two answer
options.
3. Proof — the third story shows a concrete result,
an example or a screenshot (the author names the format).
4. Call to action — the last story with one clear
action: message, tap through, save.
For every story give: the format (text/poll/screenshot),
the text length and the text itself. Tone: direct, no jargon.
When I name a topic — write the storyboard by this scheme
right away, without asking about the format.
Save it — the scheme is locked in, and the next invocation needs no format explanation.
Step 3 — call it with a single command
From then on, instead of describing the format from scratch you just name the topic and the slash command:
Make stories on the topic "AI instead of working 24/7"
/instagram-stories
Claude layers the saved instruction on top of the request and immediately returns the storyboard across all four blocks.
What you get
The output is a storyboard where every story spells out what to show, what to write and exactly where the interactive element goes. Only the specific examples and phrasing change with the topic — the block sequence (pain hook → context with a poll → proof → call to action) stays constant.
How to check the output
- The hook of the first story names a problem, not an abstract topic.
- The poll in the second story has two clear answer options, no more.
- The proof is a concrete format (a screenshot, a number, a testimonial), not a general claim.
- The last story has one action for the viewer, not several at once.
How we do it
The qvib.pro knowledge base holds more than five hundred open cards on Claude and content — including breakdowns of story formats and how to build your own skills, with no DMs and no code words. The same principle — fix the format once and call it by command — is at the core of the qvib engine: content and code are assembled along a structure defined in advance rather than from scratch on every run.
Related links
- The "Claude Skills" section of the arsenal: qvib.pro/arsenal/claude-skills
- Prompts for content: qvib.pro/arsenal/prompts-content
- How the qvib engine works: qvib.pro/