Competitor Creative Teardown

Before you start, ask me about each [input] below — one question at a time, with ready-made answers to pick from; skip whatever you can already tell.

Tear down the ads of [brand: competitor page name, as it appears in the Meta Ad Library] and tell [client: if not listed, call roasr_list_clients and let me pick] what is worth learning from it. Depth: [depth: top 3 ads through Vision (default) | top 5 ads through Vision | clusters only, no Vision].

1. **Load the RoASr method.** Call roasr_load_skill('roasr_frame') and roasr_load_skill('creative_intel') and follow them exactly.
2. **Frame the data.** Call roasr_limits, roasr_list_clients and roasr_list_accounts: plan (check that competitor intel is included), niche, margin, KPI, tone, account currencies and sync status. If currency.mixed is true, never sum money across accounts.
3. **Pull the brand's ads.** roasr_spy_search with brand set to the page name, limit 50. If nothing comes back, say so, suggest another spelling or a part of the name, and stop; never invent ads. If the page name matches several pages, list them and ask me which one.
4. **Rank by staying power.** Order by days_live and by how many variants share the same idea. days_live is a proxy: a long-running ad is a likely winner, never a confirmed one. The Ad Library shows no spend, reach or results, so write "likely" and never state the brand's budget, CPA, ROAS or targeting.
5. **Cluster.** Group the ads by angle (pain, outcome, social proof, offer, comparison, founder story), by format (video, static, carousel) and by offer type (discount, bundle, free trial, guarantee). Show for each cluster the number of ads, the median days_live and one example hook.
6. **Vision on the top ones.** Take the top ads by [depth] from distinct clusters and call roasr_vision_analyze with each ad_archive_id (adArchiveId): hook, angle, strengths, risks, and the transcript if there is one. If an ad has not been analysed, say that the analysis has to be started inside the RoASr app, and use the hook and headline from the search for that ad.
7. **Compare with the market.** roasr_spy_tag_leaderboard with limit 20: where does this brand's mix differ from the niche as a whole (a tag it over-uses or never uses)? Say whether that looks like a deliberate bet or a gap.
8. **What applies to the client.** Using niche, tone and margin from roasr_list_clients: 3 patterns worth adapting, 2 worth avoiding (and why), and 1 gap the brand leaves open. Adapt the pattern, never copy the wording. With a margin, check that any discount-led offer fits break-even ROAS = 1 / margin; without a margin, do not size offers.

**Output:** one sentence on what this brand most likely relies on and the evidence. Then a table of clusters (cluster · ads · median days_live · example hook · read) and a table of the Vision breakdowns (ad · hook · angle · strength · risk). Then 3–6 actions for the client grouped High / Medium / Nice-to-have, each with the expected effect and how to test it.

**Output rules:**
- Recommendations only — never apply changes; RoASr tools are read-only.
- Say "likely" for everything inferred from days_live; no conclusions about competitor spend or results.
- Never copy competitor wording or assets.
- If the client has no margin set, do not size offers or discounts.
- Answer in the language of this prompt.