teach-ad-breakdown
Skill0 starsChanged 27 days agoMITView on GitHub
Coach a daily direct-response rep: take one ad from the AdSwipeFile library and what the learner noticed, name the principle behind each observation, show what they missed and where the ad breaks its format's pattern, and end on one takeaway saved to their board. Use for "teach me this ad", "ad of the day", "break this ad down so I learn", or "what did I notice here".
In this repository14 other files it gives its agents.
- audit-ad-creativeSkill
- competitor-adsSkill
- creative-strategySkill
- find-winning-adsSkill
- generate-ad-conceptsSkill
- get-startedSkill
- map-competitorsSkill
- map-personasSkill
- pick-creative-mechanicSkill
- record-creative-learningSkill
- review-saved-adsSkill
- track-competitor-adsSkill
- weekly-competitor-reviewSkill
- write-hooks-from-adsSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 36
Use these headings, word for word, and drop any section you skip.
- Distinctive: line 41
What you noticed, named: only when they shared observations.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Teach an ad breakdown
Needs the AdSwipeFile connector. This skill reads the AdSwipeFile ad library of Meta ads through its MCP server. If no AdSwipeFile tools (
search_library,list_ads) are available, say so before anything else and give the step for this host. Claude Code:claude mcp add --transport http adswipefile https://adswipefile.com/api/mcp. claude.ai or ChatGPT: addhttps://adswipefile.com/api/mcpas a custom connector. Other clients: https://adswipefile.com/docs/mcp?utm_source=agent-skill&utm_medium=referral&utm_campaign=teach-ad-breakdown. Connecting opens a sign-in to an AdSwipeFile account. Steps that need only the user's own material can go on; never invent ads, brands or links in place of the library.
One ad a day, studied closely, trains the eye. This skill teaches the ad; it doesn't grade or rewrite
it (that is audit-ad-creative). The learner should leave able to spot the same move in the next ad
alone. Words are used as the creative-strategy skill's glossary defines them.
Inputs
- The ad.
get_adfor itsbreakdown(units, the first 3 seconds of speech, structure),why_it_works, tags andrun. For an ad of the learner's own, the image, frames or copy they share. Read every unit: spoken lines, on-screen words, visuals, the order. When only a frame and the copy came back, teach from those and say what you couldn't see. - What they noticed (optional), in their own half-formed words.
- The format, theirs or the ad's visual format tag.
get_formatgives its how-to-tell and how-to-build page; use it as the baseline when it exists. - No ad chosen: pick one:
list_adswithsort: "longest_running"in their niche, or the newest fromlist_followed_ads, orlist_adswith themechanicorhookthey are training on. A long run made it worth studying; say that, never that it worked.
The rep
Use these headings, word for word, and drop any section you skip. Numbered points inside each.
The format's normal pattern: name the format (both, when it mixes two) and, in a few lines, how ads in it usually run: the structure, each part's job, why it sells. This is the baseline.
What you noticed, named: only when they shared observations. One at a time, in their order: the real name of what they saw (a persuasion principle like loss aversion, a classic idea like risk reversal, a tactic like a qualifier hook), and in one or two plain sentences why it moves the viewer. Confirm what they got right before adding the rest; correct a misread kindly, with what's really happening. Give the small ones real attention.
What you missed: everything else, in the order it happens. Look at the hook (line, frame and text together), how it picks its viewer and lets others scroll past, the order of promise, proof and ask, words doing extra work, what each piece of proof proves, the offer and how it lowers risk, what the look and sound signal, and callbacks. Depth on the strongest moves beats a long list.
Where this ad breaks the pattern: where it departs from the baseline and why that was smart or what it risks. When it follows the pattern, say so and name what it does especially well.
Your one takeaway: one move to watch for, from this ad's strongest move, in fresh words each day: "Watch for ads that [move]. Ask yourself [question for the next ad]."
How it reads
Plain words, short sentences; a technical term gets one sentence of explanation the first time.
Quote the ad's real words and point to timestamps or parts of the image for every point. When you're
guessing why the brand chose something, say it's your read. Talk to the learner directly and name
the brand or viewer when you mean them. Commas, colons and full stops, no dashes. Link the ad by its url. Never state the ad's results.
Keep the rep
With the learner's go-ahead: save_ad the ad, add_ad_to_board on their reps board (create_board
"Reps" the first time), and set_ad_note with the takeaway line, so the board becomes their swipe
file of moves.
Rules
- One ad per rep. Keep calls few: the ad, and its format's page when it helps.
- A long run is a reason to study an ad, never proof it worked: write "kept running for 94 days".
Done when every observation is named, the missed moves are covered in order, the takeaway is one line, and the rep is saved where they asked.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

