Update to Unify

{
  "description": "Outreach with Unify, create email sequences, generate per-prospect copy previews, enroll prospects, and manage rep tasks (calls, LinkedIn touches, to-dos). Use when the user wants to email prospects, write outbound copy, start or edit a sequence, check what's in flight, or work their task queue.",
  "included_files": [],
  "name": "outreach",
  "skill_md_contents": "---\nname: outreach\ndescription: Outreach with Unify, create email sequences, generate per-prospect copy previews, enroll prospects, and manage rep tasks (calls, LinkedIn touches, to-dos). Use when the user wants to email prospects, write outbound copy, start or edit a sequence, check what's in flight, or work their task queue.\n---\n\n# Outreach: sequences and tasks\n\nOutreach briefs go through `run_agent` (see `agent-runs`). Sequences have a strict\nsafety lifecycle. Use its vocabulary precisely with both the agent and your user.\n\n> **Deterministic alternative.** The public API exposes direct tools for\n> inspecting and managing existing sequences, enrollments, and tasks without the\n> agent: `list_sequences` / `retrieve_sequence` / `list_sequence_steps`,\n> `list_sequence_enrollments` / `create_sequence_enrollment` /\n> `pause_sequence_enrollment` / `resume_sequence_enrollment`, and\n> `list_tasks` / `create_task` / `complete_task` / `update_task`. To export large\n> volumes of enrollments, enrollment steps, or tasks, use the async **Bulk API**\n> tools. All require an API-key connection (`X-API-Key`).\n\n## Sequence lifecycle (safety-critical)\n\n1. **Scaffold**, the reusable structure: steps (emails, calls, LinkedIn touches),\n   delays, and copy prompts. Creating a scaffold sends nothing.\n2. **Preview**: per-prospect generated copy. Still sends nothing.\n3. **Approved**: previews are ready to be enrolled.\n4. **Enrolled**: prospects are in the sequence and email actually sends.\n\nNever tell your user someone is \"enrolled\", \"started\", or that anything \"sent\"\nbefore step 4. Default briefs should end at previews: \"create previews for these\n10 people and stop for my review.\"\n\n## Writing sequence briefs\n\nInclude: who (people, by DataTable ID, List, or identities), the offer/value\nproposition and any proof points, desired tone and sender voice, step count and\nchannel mix (2–4 steps typical: email + LinkedIn + call), cadence (\"3 days\nbetween steps\"), and the sending mailbox if the user has a preference. The Unify\nagent is the copywriter: give it real context (why now, what triggered this,\nwhat the prospect cares about), not template text. If prospect context is thin,\nit will pause and ask to enrich first; that gate is correct. Relay it.\n\nPer-prospect copy edits (\"make the email to Jane mention their Series B\") edit\nthat preview only; scaffold edits propagate to everyone. Say which you mean.\n\n## Checking state\n\n\"What sequences do we have?\", \"who replied?\", \"how is sequence X performing?\",\n\"which previews are waiting on my approval?\" are all valid briefs.\n\n## Tasks\n\nReps' one-off manual to-dos: calls, LinkedIn profile views/connects/messages,\naction items. Useful briefs:\n\n- \"What tasks are due today?\" / \"what's overdue for me this week?\" (defaults to\n  the signed-in user; name a teammate to scope differently).\n- \"Create a task to call Jane Doe at Acme on Friday\" (self-assigned, manual types\n  only).\n- \"Complete/skip my ready LinkedIn tasks for Acme.\" Email-reply tasks are\n  completed by actually replying, not by marking done.\n\n## Mailbox voice profile\n\nUnify can analyze a mailbox's sent mail to build a **voice profile** that\nshapes generated copy. When a run kicks that analysis off, check progress with\n`load_mailbox_voice_profile({ connectedMailboxId })` (ID from the run result's\nstructured content), poll only while `isTerminal` is false; it's read-only.\n"
}