Update to Zime

{
  "description": "Drafts the follow-up email after one call. Recaps what was agreed and states the next step, using only what was actually said. Always a draft for you to review before sending.",
  "included_files": [],
  "name": "follow-up",
  "skill_md_contents": "---\nname: follow-up\ndescription: \"Drafts the follow-up email after one call. Recaps what was agreed and states the next step, using only what was actually said. Always a draft for you to review before sending.\"\nlicense: MIT\nmetadata:\n  zime:skill-id: 92e9d1f2-67c9-43c2-9d25-26e6cefe69d0\n  zime:tag: communication\n  zime:roles: ae,bdr\n  zime:visibility: global\n  zime:skill-version: \"1\"\n  zime:marketplace-version: 0.1.20\n---\n\n# Follow-up\n\nBefore anything else, call the Zime connector tool `skill_used` with skill_id `92e9d1f2-67c9-43c2-9d25-26e6cefe69d0`, skill_version `1` and skill_slug `follow-up`. It records which skill ran for Zime usage analytics. If the Zime connector is not available, skip this step and continue.\n\nTurns \"draft a follow-up for my Acme call\" into a sendable email. The Zime\ncall agent drafts it from what was actually said and committed, so the recap\nlines and next steps are real rather than plausible.\n\n## How It Works\n\n```\n┌─────────────────────────────────────────────────────────────────┐\n│                          FOLLOW-UP                               │\n├─────────────────────────────────────────────────────────────────┤\n│  STEP 1 — RESOLVE (this skill)                                   │\n│  ✓ list_meetings, recorded calls only                            │\n│  ✓ Ambiguous → show candidates, ask, pin call_id                 │\n├─────────────────────────────────────────────────────────────────┤\n│  STEP 2 — DELEGATE (Zime agent)                                  │\n│  + ask_zime drafts from real commitments and next steps, naming the call │\n│  + Returns the draft; this skill formats and captions it         │\n├─────────────────────────────────────────────────────────────────┤\n│  ALWAYS                                                          │\n│  ! Delivered as a DRAFT — the rep sends it, not this skill       │\n├─────────────────────────────────────────────────────────────────┤\n│  LOCAL FALLBACK (no zime-mcp)                                    │\n│  ~ Draft from a transcript file the user provides                │\n└─────────────────────────────────────────────────────────────────┘\n```\n\n## Usage\n\n```\n@follow-up <company or call> [+ when]\n```\n\nIn the ChatGPT desktop app you can also type `/follow-up`.\n\nDraft the follow-up for the call the user named.\n\n## Routing\n\n- A structured internal summary rather than a customer email → `review-call`.\n- Just the open items and owners → `actions-commitments`.\n- Prep for the NEXT call → `call-prep`.\n\n## What I Need From You\n\nWhich call. If the user wants a particular emphasis (\"keep it short\", \"push\nfor the pilot\", \"address the pricing concern\"), say so — that shapes the\ndraft. Only recorded calls can be drafted from.\n\n## MCP mode (required when zime-mcp is connected)\n\n**Required tools:** `list_meetings` (resolve) and `ask_zime` (draft).\n\n> `ask_zime` routes to Zime's global agent, with no separate `call_id`\n> argument — name the call (title and date) in the question text. Writing\n> the email yourself while it's available is a failure of this skill: the\n> agent knows which commitments were actually made, and an invented promise\n> in a customer-facing email is the most expensive mistake this skill can\n> make.\n\n### Step 1 — resolve the call\n\n```json\n{ \"query\": \"Acme\", \"start_date\": \"2026-08-06\", \"end_date\": \"2026-08-13\", \"recorded\": true }\n```\n\n`multiple_matches` → show candidates with dates, ask, pin. `no_match` → say\nthe window likely excludes it or it wasn't recorded; offer to widen. Never\ndraft from a different call.\n\n### Step 2 — delegate the draft\n\n```json\n{ \"question\": \"Draft a follow-up email to the customer for the Acme call on Aug 12: recap what was discussed, restate the commitments we made, and propose the agreed next step with timing.\" }\n```\n\nAdd the user's emphasis to the question verbatim if they gave one. Resolve\npronouns to real names — the agent has no memory of this conversation.\n\n### Outcomes\n\n- **A draft** — deliver per Output below.\n- **The agent has nothing** (e.g. no commitments detected) — say so and don't\n  manufacture next steps to fill the email.\n- **An error** — `{\"error\": \"<CODE>\", \"message\": \"<sentence>\"}`. Show the\n  user the message. `INTERNAL_ERROR` retry once; `TIMEOUT` retry once;\n  `CREDIT_LIMIT` stop and relay the message; `UNAUTHORIZED` / `FORBIDDEN`\n  means reconnect or lack of access. If it fails twice, say plainly the\n  drafting service couldn't be reached and offer the local fallback — never\n  pass a hand-written email off as agent-backed.\n\n## Output\n\nTwo parts: a one-line provenance caption, then the email in plain text.\n\n```markdown\n**Draft follow-up — [call title]** · [date] · to [recipients]\n_Review before sending. Every commitment below came from the call itself._\n\n---\n\nSubject: [subject line]\n\nHi [name],\n\n[recap of what was discussed]\n\n[commitments made, restated plainly]\n\n[the agreed next step, with timing]\n\nBest,\n[rep name]\n```\n\n### Email style rules\n\nThe body is going into an email client, so:\n\n1. **No markdown formatting.** No asterisks, no bold, no headers. Plain text\n   that reads naturally anywhere.\n2. **Short paragraphs**, blank line between sections.\n3. **Plain dashes** for lists, not bullets or fancy formatting.\n4. **Concise but complete** — customers are busy; don't pad.\n\n**Good:**\n```\nHere's what we agreed:\n- Quote for 20 seats by Friday\n- Security review doc from your side\n- Follow-up call the week of the 25th\n```\n\n**Bad:**\n```\n**What We Agreed:**\n- **Quote** for 20 seats by Friday\n```\n\n### Grounding rules\n\n- Every commitment, number, date, and name in the draft must have come from\n  the agent's output. If the agent didn't return it, it doesn't go in the\n  email — a plausible-sounding promise to a customer is worse than an\n  incomplete draft.\n- If a needed piece is missing (e.g. no date was agreed), leave a visible\n  `[confirm timing]` placeholder rather than inventing one.\n- Never state a discount, price, or contractual term the call didn't contain.\n- Always caption it as a draft. This skill does not send email.\n\n### Render it inline in the chat\n\nDeliver the caption and the email **inline in the chat response**, as plain\ntext. The rep is going to select the email body and paste it into a mail\nclient, so keep it in the chat rather than in a separate document or panel.\n\nThe email style rules above still govern the body — plain text, no markdown,\nplain dashes for lists. Keep any `[confirm timing]` placeholders visible\nrather than tidying them away.\n\n## Tips\n\n1. **Say the emphasis you want** — \"short and push for the pilot\" changes the\n   draft meaningfully.\n2. **Placeholders are a feature** — `[confirm timing]` beats a made-up date.\n3. **Read before you send** — the draft is grounded, not proofread for tone\n   against your relationship with this customer.\n\n## Local mode (only when no zime-mcp server is connected)\n\nIf the user provides a transcript, draft from that file only, with the same\ngrounding rules and the same draft caption. Open with one line saying the\ndraft covers only the provided transcript.\n\n## What this sends where\n\nMCP mode sends the query words and date range (to `list_meetings`), then the\ndrafting question naming the call (to `ask_zime`). Local mode reads\nonly the provided file. Neither mode sends email. At the start, the skill\ncalls the Zime connector tool `skill_used`, which records the skill's id,\nversion and name with your user and session in Zime's usage log.\n\n## Related Skills\n\n- **review-call** — the internal summary version\n- **actions-commitments** — what's still open across calls\n"
}