agentleFS
adswipefileadswipefile/adswipefile-plugin/skills/map-personas/SKILL.md

map-personas

Skill0 starsChanged 27 days agoMITView on GitHub

Find every persona in a product's customer reviews (who they are, the words they call themselves, variations, overlaps and filmable trigger moments), then pull reference ads for each persona from the AdSwipeFile ad library. Use when someone shares reviews and wants personas, asks who their customers are, wants customer avatars or ICP segments for ads, or a concept needs a persona they lack.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep the reviews that tell a positive story: personas come from happy customers.

    line 33
  2. Distinctive:

    Primary persona: when the largest has at least twice the members of the next.

    line 58
  3. Very specific:

    Read the closest few together with get_ads (up to 25 per call).

    line 82
  4. Distinctive:

    Never invent a persona, a moment or a quote the records do not carry.

    line 105
  5. Distinctive:

    Never state an ad's results: spend, ROAS, conversions, or that one ad beat another.

    line 106

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Map personas

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: add https://adswipefile.com/api/mcp as a custom connector. Other clients: https://adswipefile.com/docs/mcp?utm_source=agent-skill&utm_medium=referral&utm_campaign=map-personas. 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.

Find the people an ad can be made for, from what happy customers wrote about themselves. Words (persona, trigger moment, awareness stage) are used as the creative-strategy skill's glossary defines them.

The reviews come from the user: pasted, a file, a page they link, or another connector they have. AdSwipeFile holds ads, not reviews. Reviews of a competitor's product stay labelled as that competitor's customers in every output.

1. Intake

  1. One product per read. Split a mixed set by product; ask when a review's product is unclear.
  2. Drop duplicates (same reviewer, same text). Count what is left.
  3. Incentivized reviews ("received free", "in exchange for"): count them and set them aside unless the user asks to keep them.
  4. Score each review 1 to 5 for signal: 1 is three words or fewer or says nothing about the person ("love it"), 5 is a vivid paragraph about their life. Drop the 1s.
  5. Keep the reviews that tell a positive story: personas come from happy customers. A mixed review keeps its positive half.
  6. Floor: about 50 analyzed reviews. Under that, report the counts and say the read is thin; go on only if the user wants a rough read, and mark every persona as thin.

2. One record per customer

For each analyzed review write a record from that review alone: id; who they are (life context); what they call themselves, verbatim ("I'm a night-shift nurse"); any second identity; the one thing they use it for; pains from the problem; pains from their old fix; what they wanted; what they tried; their hesitation; what made them buy; what changed; when they reach for it now; where, when and who; their best lines. Leave a field empty when the review is silent. Every number below is counted from these records.

3. Group records into personas

  • A persona is an identity someone would claim, or a thing they regularly do ("after my run"). A pain, a habit or a buying behaviour is a field inside a persona. A record with no identity stays unassigned; count them.
  • The story decides the split. Two groups are separate personas when an ad would tell them different stories; variations of one persona when only the details change. Name the close calls.
  • Buying for someone else (a partner, a parent, a child) is one persona with each relationship as a variation, until one relationship has enough members to stand alone.
  • A customer belongs to the persona their story is built around; a second identity is an overlap, recorded on both ("runner who travels for work, 3").
  • Primary persona when the largest has at least twice the members of the next. Otherwise report all as equals and say so.

Report every persona, largest first, no minimum size. Each block, from its own members' records: name (the identity), members and share, two sentences on who they are, roles verbatim with counts, variations, overlaps, use case, both pain layers with counts, wanted, tried, hesitation, what made them buy, what changed, usage moments, how they talk, 3 to 5 quotes from different members. Then shared pains: each pain carried by more than one persona, with counts per persona.

4. Trigger moments, per persona

From each persona's own members, list every trigger moment, most common first: the moment in one filmable line, members, the stage it moves them to, their role in that moment, setting and who is there, feeling in their words, need versus want, where the product enters, and the review lines behind it. A moment needs a where, a when and what happened. When the reviews carry no situational depth, say so: that is a finding, and inventing a moment is the failure this step exists to prevent.

5. Reference ads per persona

For each persona, find 2 to 3 ads that stage its moments. Keep calls few.

  1. search_library with the moment's plain words, 1 to 3 of them ("night shift", "gym bag", "new mom"). It also gives the niche slug.
  2. list_ads with that niche and sort: "longest_running", limit 10 to 20.
  3. Read the closest few together with get_ads (up to 25 per call). Keep the ones whose breakdown or copy shows this persona's identity or moment.

For each, give its url, the moment it stages and how it is built (from why_it_works and breakdown). A long run is a reason to study an ad: write "kept running for 94 days", never that it worked.

6. Hand it over

The report goes in the reply (or a file, if the client can write files): header with brand, product, source (own brand or "competitor: name"), counts and date; the personas; the trigger moments; the reference ads. Then offer to keep the references on AdSwipeFile, one board per persona: create_board, add_ad_to_board, and set_ad_note naming the moment each ad shows, in the client's workspace when the user works for several. Only on a yes. Then offer share_board, so the personas' references reach the client or team as one link.

Done when every analyzed record is assigned or counted as unassigned, every persona and trigger moment cites record ids, and each persona has its reference ads linked. Concepts from here: generate-ad-concepts.

Rules

  • Quotes are exact substrings of a review. Your summaries never go in quotes or "how they talk".
  • Never invent a persona, a moment or a quote the records do not carry.
  • Never state an ad's results: spend, ROAS, conversions, or that one ad beat another.
  • A persona is who an ad is made for, not a demographic and not who an ad reached.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.