agentleFS
EricStrohmaierEricStrohmaier/adchat-mcp/skills/build-audiences/SKILL.md

build-audiences

Skill1 starsChanged 3 days agoMITView on GitHub

Create Meta (Facebook/Instagram) audiences through AdChat: website visitors and purchasers (pixel), page and Instagram engagers, lookalikes, interest targeting, and audience size estimates. Use when the user asks for retargeting, a lookalike, a custom audience, interest targeting, exclusions, or "who should I target".

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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
  1. Distinctive:

    Always exclude recent purchasers from prospecting and retargeting unless the user sells repeat purchases on purpose.

    line 64

≈ “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

Build audiences

Goal: give the user the few audiences that matter for their stage, named clearly, and used correctly (including exclusions). For most small accounts with a working pixel: broad prospecting, a lookalike of buyers, and a retargeting audience are enough.

Pick the right audience

NeedAudienceHow
Bring back site visitorsWebsite audience, 30 or 180 daysmeta_audience, action: create_website
Bring back buyers / exclude buyersWebsite audience on the Purchase eventcreate_website with include_events
Bring back people who engagedPage or Instagram engagersmeta_audience, action: create_engagement
Find people like buyersLookalike 1% of purchasersmeta_audience, action: create_lookalike
Cold, no data yetBroad or interestsmeta_search_targeting, kind: interests

Steps

  1. Context. get_account_briefing. See what exists: meta_list with level: custom_audiences. Don't build duplicates.
  2. Pixel. meta_list with level: pixels and meta_diagnose with kind: pixel_health. Website audiences need a pixel that is firing the events.
  3. Create (one at a time, tell the user each one):
    • Website: meta_audience with action: create_website, pixel_id, name, retention_days (max 180), optional url_contains, include_events, exclude_events.
    • Engagement: action: create_engagement, source_type (page or instagram), source_id (from meta_get_account with include: ["pages"]), event, retention_days.
    • Lookalike: action: create_lookalike, origin_audience_id (a source with at least ~100 people in the country, ideally 1,000+), country, ratio (0.01 = 1%).
  4. Interests (cold). meta_search_targeting with kind: interests and kind: interest_suggestions. Keep 3–6 related interests in one group.
  5. Size check. meta_estimate_audience with the full targeting. Retargeting under ~1,000 people barely delivers; prospecting under ~500k is narrow.
  6. Use it. Audiences do nothing until an ad set uses them. Offer to add them to a new PAUSED ad set (launch-campaign) or to an existing one with meta_update (level: adset, new targeting); editing an ACTIVE ad set needs a Yes.

What to show the user

Created (not used by any ad set yet)
- Website visitors 30d (excl. purchasers)   · pixel Acme  · fills over ~24h
- Purchasers 180d                           · source for lookalike
- Lookalike 1% US – Purchasers 180d         · ~2.3M people
Suggested use: Prospecting = Lookalike 1% US, exclude Purchasers 180d.
Retargeting = Website visitors 30d, exclude Purchasers 180d.

Guardrails

  • Creating audiences needs no Yes on autopilot (it doesn't spend), but say what you created. Deleting an audience (meta_delete_audience) can't be undone: always ask first.
  • Name audiences by source, window and country: Purchasers 180d, LAL 1% US – Purchasers 180d.
  • New audiences take hours to fill; lookalikes can take up to a day.
  • Special ad categories (housing, credit, employment, politics) restrict targeting and lookalikes. Check the campaign before suggesting.
  • Always exclude recent purchasers from prospecting and retargeting unless the user sells repeat purchases on purpose.

Example prompt

"Set up retargeting for people who visited my site but didn't buy, and a lookalike of my customers in Canada."

Briefing → existing audiences → pixel health → website visitors 30d excluding Purchase → Purchasers 180d → lookalike 1% CA → size estimate → suggested use → offer to add them to ad sets.

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.