Update to august

{
  "name": "august-interview",
  "description": "Run a structured 1:1 interview with a single August twin off a discussion guide (a list of questions or topics the user provides), and return a clean synthesized write-up instead of raw back-and-forth. Use when the user wants depth on one persona and does NOT want to read the full transcript themselves. Triggers on: \"interview a twin\", \"structured interview\", \"run a discussion guide\", \"walk a twin through these questions\", \"1:1 twin interview\", \"/august-interview\".",
  "included_files": [],
  "skill_md_contents": "---\nname: august-interview\ndescription: >\n  Run a structured 1:1 interview with a single August twin off a discussion\n  guide (a list of questions or topics the user provides), and return a\n  clean synthesized write-up instead of raw back-and-forth. Use when the\n  user wants depth on one persona and does NOT want to read the full\n  transcript themselves. Triggers on: \"interview a twin\", \"structured\n  interview\", \"run a discussion guide\", \"walk a twin through these\n  questions\", \"1:1 twin interview\", \"/august-interview\".\n---\n\n# August Interview\n\nDrive a structured 1:1 interview with a single August twin using a\ndiscussion guide, then hand back a clean write-up. Tools live under\nthe August MCP.\n\n**When to use:**\n- User has a list of questions/topics they want walked through with one twin\n- User wants depth on a single persona (not a range of reactions)\n- User explicitly does not want to read the full transcript\n\n**When NOT to use:**\n- User wants varied reactions across personas → use focus group tooling\n- User wants a single gut-check reaction → use `/august-feedback`\n- User just wants to find the right twin → use `list_organizations` and\n  `list_twins` directly\n\n**Core rule:** if org, twin, or the discussion guide itself is unclear, ASK\nbefore calling any tool. Better one clarifying question than an interview\nwith the wrong persona off the wrong questions.\n\n---\n\n## STEP 1 — Confirm the discussion guide\n\nBefore touching August, make sure you have:\n\n- A concrete list of questions or topics (the \"discussion guide\")\n- A rough sense of interview depth per question (one probe, or several\n  follow-ups?)\n- The intended audience for the twin (which org / persona type)\n\nIf the user pasted a wall of text, extract the questions into an ordered\nlist and mirror it back for confirmation. If the guide is vague (\"ask them\nabout pricing\"), ASK for the actual questions — do not invent them.\n\n---\n\n## STEP 2 — Pick the organization\n\nThe August MCP is org-scoped. Every twin belongs to an org.\n\n1. If the user named a brand or audience (GNC, Backcountry, Fermat\n   dashboard user), match to an org via `list_organizations`.\n2. If they haven't, infer from the discussion guide: consumer-facing\n   questions → brand org; internal/dashboard/workflow questions → Fermat\n   org.\n3. If ambiguous, ask the user with 2–4 candidate orgs pulled from\n   `list_organizations`. Do not guess.\n\n---\n\n## STEP 3 — Pick the specific twin\n\nCall `list_twins` for the chosen org.\n\n- Match twin traits to the discussion guide's target persona.\n- If exactly one twin clearly fits, proceed.\n- If 2+ twins reasonably fit, ask the user with the top 2–3\n  candidates and one-line summaries. Never silently pick.\n\nRecord the selected `twin_id` for the interview call.\n\n---\n\n## STEP 4 — Start the interview and loop the guide\n\n1. Call `start_interview` with the selected `twin_id` and save the returned\n   `interview_id`.\n2. For each question in the discussion guide, call `ask_question` with that\n   `interview_id`. Include the discussion framing in the first question and\n   preserve the guide order.\n3. If a response is thin or dodges the question, follow up with ONE\n   clarifying `ask_question` before moving on. Do not spiral — the goal\n   is coverage of the full guide, not exhaustive probing on one item.\n4. Once every guide question has been asked (and any needed follow-ups are\n   done), call `get_interview_transcript` with the interview ID as\n   `conversation_id` to retrieve the full record.\n\nPass the actual artifact (copy, screenshot description, product details)\nwhen a question references one. Vague prompts produce vague answers.\n\n---\n\n## STEP 5 — Synthesize the write-up\n\nDo NOT return the raw transcript. Return this structured format:\n\n```\n# Interview write-up — <twin name>, <org>\n\n## TL;DR\n2–3 sentence overall read on the twin's stance across the guide.\n\n## Per-question responses\nFor each question in the original guide order:\n\n### Q1 — <the question, verbatim>\n- **Summary:** 1–3 sentences capturing the twin's answer.\n- **Key points:** 2–4 bullets of the specific things they said.\n- **Confidence / hedging:** note if they hesitated, hedged, or were firm.\n\n(Repeat for each guide question.)\n\n## Cross-cutting themes\nPatterns that showed up across multiple questions (e.g., recurring\nconcern about price, recurring positive on brand trust).\n\n## Surprises\nAnything the twin said that you did not expect given the persona\nprofile — worth flagging to the user.\n\n## Verbatim quotes\n3–6 short direct quotes from the transcript, attributed with the\nquestion they came from. Preserve tone and phrasing exactly.\n\n## Source\nTwin name + twin_id + org + interview id, so the user can trace back.\n```\n\nIf the transcript came back thin, off-topic, or the twin refused\nquestions, say so explicitly at the top of the write-up and offer to\nre-route to a different twin.\n\n---\n\n## Quick reference — when to ask vs. proceed\n\n| Situation | Action |\n|---|---|\n| Discussion guide is concrete | Proceed |\n| Discussion guide is vague (\"ask about X\") | **ASK** for the actual questions |\n| Org is named or obvious | Proceed |\n| Org is unclear | **ASK** from `list_organizations` |\n| One twin clearly fits | Proceed |\n| 2+ twins fit | **ASK** with top 2–3 candidates |\n| Twin dodges one question | Follow up once, then move on |\n| Twin dodges most questions | Stop, flag, offer to re-route |\n"
}