Update to august

{
  "name": "august-focus-group-report",
  "description": "Take an already-finished August focus group and turn its raw results into an actual readout — consensus points, disagreements with named sides, standout quotes, a rough sentiment split, and unresolved follow-up questions. Post-hoc synthesis only; does NOT trigger a new focus group. Use when the user has an FG that already ran and wants it read out for them instead of scrolling raw output. Triggers on: \"focus group report\", \"synthesize focus group\", \"read out this focus group\", \"FG readout\", \"summarize the focus group\", \"/august-focus-group-report\".",
  "included_files": [],
  "skill_md_contents": "---\nname: august-focus-group-report\ndescription: >\n  Take an already-finished August focus group and turn its raw results into\n  an actual readout — consensus points, disagreements with named sides,\n  standout quotes, a rough sentiment split, and unresolved follow-up\n  questions. Post-hoc synthesis only; does NOT trigger a new focus group.\n  Use when the user has an FG that already ran and wants it read out for\n  them instead of scrolling raw output. Triggers on: \"focus group report\",\n  \"synthesize focus group\", \"read out this focus group\", \"FG readout\",\n  \"summarize the focus group\", \"/august-focus-group-report\".\n---\n\n# August Focus Group Report\n\nPost-hoc synthesis of a completed August focus group. This skill does\nNOT run a new focus group — it reads out an existing one. Tools live\nunder the August MCP.\n\n**When to use:**\n- A focus group has already been run (the user references it by name /\n  id, or says \"the focus group I did yesterday\")\n- The user wants consensus, disagreements, and quotes — not raw output\n- The user is preparing a share-out, doc, or decision brief off the FG\n\n**When NOT to use:**\n- The user wants to START a new focus group — that's a different flow\n  (`start_focus_group` + `run_focus_group_round`), not this skill\n- The user wants deep 1:1 depth on one twin → use `/august-interview`\n- The user wants a quick gut check → use `/august-feedback`\n\n**Core rule:** never fabricate consensus or attribution. If the raw\nresults don't clearly show who said what, say so — do not paper over\ngaps to make the readout cleaner.\n\n---\n\n## STEP 1 — Identify the focus group\n\nGet to a specific `focus_group_id` before doing anything else.\n\n1. If the user provided an id or exact name, use it directly.\n2. If they described the FG loosely (\"the pricing one from last week\",\n   \"the Backcountry hero image FG\"), call `list_focus_groups` on the\n   likely org and match by name/date/topic.\n3. If multiple candidates match, ask the user with the top 2–3\n   with one-line summaries (topic + date + twin count) and let the user\n   pick. Never guess.\n4. If no FG matches, tell the user plainly — do not fabricate results,\n   and do not silently start a new one.\n\n---\n\n## STEP 2 — Pull the raw results\n\nCall `get_focus_group_results` with the confirmed `focus_group_id`.\n\nAlso note, from the results payload:\n- List of participating twins (names + ids)\n- List of stimuli that were shown\n- Number of rounds run\n\nYou'll need all of these for attribution in the readout.\n\nIf the results are empty or the FG hasn't finished, stop and tell the\nuser — do not synthesize a partial FG as if it were complete.\n\n---\n\n## STEP 3 — Synthesize\n\nRead the full results carefully. Build the readout in this order:\n\n1. **Group by stimulus / question**, not by twin. The unit of analysis\n   is \"what did the group think about X?\" not \"what did Twin A think?\"\n2. **For each stimulus, sort takes into agree / disagree / mixed** based\n   on the actual content — do not force a 3-way split if there wasn't\n   one.\n3. **Attribute every disagreement** to the specific twins on each side.\n   Attribution is the whole point of running a group vs. a single\n   interview.\n4. **Extract quotes** that are punchy, specific, and representative —\n   not the longest, the shortest, or the most anodyne.\n5. **Score sentiment** at the twin level per stimulus (positive / mixed\n   / negative). Report the split as raw counts, not percentages — the\n   sample is too small for percentages to be honest.\n6. **Flag unresolved questions** — places where twins clearly wanted\n   more info, contradicted themselves, or where the stimulus didn't\n   land clearly enough to draw a conclusion.\n\n---\n\n## STEP 4 — Return the readout\n\nUse this exact structure:\n\n```\n# Focus group readout — <FG name>, <org>\nParticipants: <N twins> · Stimuli: <M> · Rounds: <R>\n\n## TL;DR\n3–5 sentences capturing the overall read. What did the group broadly\nbelieve? Where was there real disagreement? What's the biggest signal\nfor the decision this FG was meant to inform?\n\n## Consensus — what the group agreed on\nBulleted list. Each bullet: one claim, then in parens the count of\ntwins who supported it (e.g., \"Price feels too high (5 of 6)\"). Only\ninclude items with clear majority backing.\n\n## Disagreements — where the group split\nFor each real split:\n- **The question / stimulus:** <what they were reacting to>\n- **Side A (<count>):** <twin names> — <their position, 1–2 sentences>\n- **Side B (<count>):** <twin names> — <their position, 1–2 sentences>\n- **Why the split matters:** 1 sentence on the decision implication.\n\n## Sentiment breakdown\nPer stimulus, raw counts:\n- Stimulus 1: <N positive> / <N mixed> / <N negative>\n- Stimulus 2: ...\n(Only include stimuli where sentiment was actually captured.)\n\n## Standout quotes\n3–5 direct quotes, each attributed to a specific twin and stimulus.\nPreserve the twin's voice and phrasing.\n> \"quote\" — <twin name>, on <stimulus>\n\n## Unresolved — worth a follow-up\nBulleted list of questions the FG raised but didn't answer. These are\ncandidates for a targeted follow-up interview via `/august-interview`.\n\n## Source\nfocus_group_id, org, participants list (names + ids), so the user can\ntrace back.\n```\n\nIf the raw results are thin (short answers, low engagement, twins\nstaying vague), say so at the top and note that a re-run with tighter\nstimuli might help — do not inflate weak signal into strong claims.\n\n---\n\n## Quick reference — when to ask vs. proceed\n\n| Situation | Action |\n|---|---|\n| User gave an FG id or exact name | Proceed |\n| User described the FG loosely, one clear match | Proceed |\n| Multiple FGs match description | **ASK** with top candidates |\n| No FG matches | Tell the user; do not fabricate; do not start a new one |\n| FG has finished, results are rich | Full readout |\n| FG has finished but results are thin | Deliver readout + honest thinness note |\n| FG hasn't finished / no results | Stop; report status back |\n"
}