Update to august

{
  "name": "august-concept-compare",
  "description": "Run the same concept, design, or copy past August twins across two or more variants (A/B/C/D) and return a side-by-side comparison instead of one twin's take at a time. Use whenever the user has multiple candidate concepts and wants twin-simulated preference signal before committing — landing page variants, packaging concepts, positioning angles, hero designs, pricing framings, ad creatives. Different from /august-feedback (single artifact). Also handles text-only variants (headlines, CTAs, taglines) — hard-scope output to phrase-level notes in that case. Triggers on: \"compare these concepts\", \"A/B test with twins\", \"which variant do twins prefer\", \"run these variants past twins\", \"concept test\", \"/august-concept-compare\".",
  "included_files": [],
  "skill_md_contents": "---\nname: august-concept-compare\ndescription: >\n  Run the same concept, design, or copy past August twins across two or more\n  variants (A/B/C/D) and return a side-by-side comparison instead of one\n  twin's take at a time. Use whenever the user has multiple candidate\n  concepts and wants twin-simulated preference signal before committing —\n  landing page variants, packaging concepts, positioning angles, hero\n  designs, pricing framings, ad creatives. Different from /august-feedback\n  (single artifact). Also handles text-only variants (headlines, CTAs,\n  taglines) — hard-scope output to phrase-level notes in that case. Triggers\n  on: \"compare these concepts\", \"A/B test with twins\", \"which variant do\n  twins prefer\", \"run these variants past twins\", \"concept test\",\n  \"/august-concept-compare\".\n---\n\n# August Concept Compare\n\nCompare 2–4 variants of the same concept across August twins and return a\nmatrix of reactions plus a recommended winner. Tools live under\nthe August MCP.\n\n**When to use:** the user has multiple candidate concepts and wants relative\npreference signal (which is stronger, and why).\n\n**When NOT to use:**\n- Only one concept exists → use `/august-feedback` instead.\n- User wants step-by-step friction on a flow → use `/august-flow-audit`.\n\n**Text-only variants (headlines, CTAs, taglines, body copy):** still use this\nskill, but hard-scope the output to phrase-level notes — comment only on\nwording, not on layout, imagery, or design drift. Flag any design/layout\nobservations that surface anyway as \"out of scope — worth a full concept\ntest?\" rather than including them in the main matrix.\n\n**Core rule:** if the org, the audience, or the comparison format is\nunclear, ASK once. Don't silently pick — mis-routed comparisons are worse\nthan mis-routed single feedback because they influence a choice.\n\n---\n\n## STEP 1 — Confirm the variants + audience\n\nBefore any MCP call, be clear on:\n\n- **How many variants?** (2, 3, or 4 — cap at 4; more than that dilutes\n  signal.)\n- **What form?** Text, image/screenshot, Figma link, full page mock?\n- **What are they variants OF?** A hero section, an entire landing page,\n  a pricing card, a value prop, a packaging design, etc.\n- **What audience should judge?** Brand shopper twins (consumer artifact)\n  or Fermat-org dashboard-user twins (internal tool)?\n\nIf any of these are missing, ASK. Ask the user for closed choices\n(2–4 options); prose otherwise.\n\n---\n\n## STEP 2 — Pick the organization\n\nCall `list_organizations`. Then match audience → org:\n\n| Artifact | Likely org |\n|---|---|\n| Consumer landing page, DPP, ad, packaging, pricing | Brand org (GNC, Backcountry, etc.) |\n| Fermat/August dashboard UI, admin tooling, internal workflow | Fermat org |\n\nIf the user named a brand or audience, resolve directly. If ambiguous, ASK\nwith `list_organizations` results as options.\n\n---\n\n## STEP 3 — Pick the comparison format\n\nThree viable shapes. Default to (B) unless the user signals otherwise.\n\n**(A) Single twin sees all variants (one interview)**\n- Best for: tight cost, quick gut check, one strong-fit persona.\n- Method: pick 1 twin via `list_twins`, call `start_interview(twin_id)`, then\n  call `ask_question(interview_id)` once per variant with matching framing\n  and the artifact in `design_context`. Do NOT reveal that other variants\n  exist until the final preference question.\n\n**(B) One focus group sees all variants together — DEFAULT**\n- Best for: real A/B/C preference signal with diversity.\n- Method: use `list_focus_groups` to select an existing ready group, then\n  `start_focus_group`. Attach Figma variants with `add_focus_group_stimuli`;\n  for text or described-image variants, include the labelled artifacts in\n  the round question instead. Run the reaction round, then a preference\n  round, then `end_focus_group` → `get_focus_group_results`. If no suitable\n  ready group exists, ask the user to configure one in August or use (A).\n\n**(C) One focus group per variant (parallel groups)**\n- Best for: when priming across variants would contaminate reactions\n  (e.g., strongly branded concepts where seeing variant B changes how you\n  read variant A).\n- Method: use N existing ready groups with matched twin composition, one per\n  variant. The MCP cannot create groups or select their panels. If those\n  groups do not exist, use (A) or ask the user to configure them in August.\n\nIf the user didn't specify, ASK which shape — offer A/B/C as options.\n\n---\n\n## STEP 4 — Run it\n\nFor each MCP call, pass the actual artifact content — the copy text, image\ndescription, Figma frame contents, or pasted screenshot description.\nVague stimuli = vague comparison.\n\n**Prompts to use in the group / interview:**\n1. Reaction round: \"Here are [N] variants of [artifact]. React to each one\n   independently: what stands out, what confuses you, what makes you want\n   to click / keep reading / bounce.\"\n2. Preference round: \"Now pick the one that would most make you [target\n   action — buy, sign up, keep browsing]. Explain why in one sentence.\"\n\nRetrieve results via `get_focus_group_results` or\n`get_interview_transcript` depending on the route.\n\n---\n\n## STEP 5 — Synthesize into a side-by-side matrix\n\nDo NOT dump raw transcripts. Return this template:\n\n```\n## Concept Comparison: [artifact name]\nAudience: [org] · Format: [single twin / focus group / N parallel groups]\nTwins: [names or count]\n\n| Variant | What landed | What didn't | Who preferred it | Why |\n|---|---|---|---|---|\n| A — [label] | ... | ... | [twin names or count] | [core reason] |\n| B — [label] | ... | ... | ... | ... |\n| C — [label] | ... | ... | ... | ... |\n\n### Recommended winner: **Variant [X]**\n- **Why it wins:** [1–2 sentence synthesis of the strongest signal]\n- **Tradeoff to know:** [what it loses vs. runner-up — every winner has one]\n- **Suggested next move:** [ship as-is / merge X's headline + Y's visual /\n  re-test with tighter audience / etc.]\n\n### Source\n[Focus group name(s) or twin name(s), org, dates]\n```\n\nIf signal is thin, tied, or contradictory across twins, SAY SO explicitly\nin the \"Recommended winner\" section — don't force a call. Offer to re-run\nwith a different format (e.g., swap to parallel groups) or a different\naudience.\n\n---\n\n## Quick reference\n\n| Situation | Action |\n|---|---|\n| 2–4 concrete variants + clear audience | Proceed with default (B) |\n| Only 1 variant | Redirect to `/august-feedback` |\n| Text-only variants | Proceed, but hard-scope output to phrase-level notes |\n| Variants strongly primed / branded | Ask about (C) parallel groups |\n| Org or audience unclear | ASK once, then proceed |\n"
}