The per-call runbook for turning a finished sales call's transcript — found in the connected Google Drive folder — into an updated HubSpot deal (or a flagged handoff when no deal matches), filed follow-up tasks in {{linear_team}}, and a drafted recap email — held for the rep's approval before it sends.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
The call recorder drops the transcript into the connected Google Drive folder
({{transcript_folder}}) once a call finishes. Open the new file via
google_drive and read its full content — every question raised, every
answer given, every commitment either side made. This is the entire input for
the session; no other call's context loads in.
Identify the deal from the call's attendees (contact email domain, company name).
| Signal | Routing |
|---|---|
| Exactly one deal matches confidently | Pull its current stage, notes, and contact list before writing anything, so Step 4's update is a diff against reality, not a guess |
| No deal matches, or more than one plausibly matches | Do not guess — skip Step 4 and flag the call instead when you hold it for approval (Step 7), so a human links it to the right deal manually |
Read the shape of a good recap, how the team phrases next steps, which kinds of objections need a considered answer, and this account's own history (past calls, open threads) from memory. A recap that revisits an earlier commitment should reference it, not restate it as new.
Skip this step entirely if Step 2 didn't produce a confident deal match — go straight to Step 5, and let Step 7 carry the flag instead. Otherwise write back what the call actually covered:
| Field | Update |
|---|---|
| Stage | Advance only if the call clearly moved the deal (verbal commit, next step scheduled) — otherwise leave it |
| Notes | Append a summary of the call: what was discussed, decided, and asked |
| Contacts | Add any new attendee who isn't already on the deal |
Create one Linear task per commitment in {{linear_team}}: title is the commitment phrased as a task, description is the context plus a link back to the call, due date from anything the call implied ("by Friday"), assignee is the rep who hosted the call (the owner/host field on the transcript file or recording). If that isn't resolvable, file the task unassigned with a note instead of guessing.
Write the recap to the account's history and the team's playbook: a short summary of the call, a direct answer to every open question that came up, and the next steps restated plainly. Ground it in what was actually said — never a generic template with the account name swapped in.
Place the drafted recap in {{approval_channel}} as a draft only. Do not send it. If Step 2 found no confident deal match, put a flag at the top of the held item instead of a deal reference: which call, why no deal matched (no match / multiple plausible matches), and a link to the transcript, so a human can link it to the right deal manually — the deal update stays pending until that link exists. Otherwise, the deal update and the filed tasks are already live by this point — only the outbound email waits on the rep.
</workflow> <guardrails> - **Never send.** The recap email is always a draft held in {{approval_channel}} for the rep to review, edit, and send. - **One call, one session.** Nothing carries over between calls; each is scoped entirely to its own transcript and deal. - **Flag instead of guess.** A call that doesn't confidently match exactly one HubSpot deal skips the CRM update and is held in {{approval_channel}} with the ambiguity noted, instead of updating the wrong deal or silently skipping the call. - **CRM writes are grounded in the transcript.** Stage advances, notes, and contacts only reflect what the call actually covered — never inferred beyond it. - **Scoped secrets.** Google Drive, HubSpot, and Linear credentials are injected by the connector at runtime, scoped to this agent's grant. - **No chat posts.** The HubSpot deal, Linear, and the held email draft are the only outputs — this skill does not post to Slack or any other channel. </guardrails> </skill>まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Policy runbook for turning a Slack access request into a least-privilege, policy-checked GitHub or AWS IAM grant. Covers the role-to-grant mapping, extra-scrutiny cases, the approval handshake, and how an applied grant gets logged.
日本語の概要は準備中です。原文の説明を表示しています。
Get a complete picture of any company or person before outreach. This skill always works with web search, and gets significantly better with enrichment and CRM data.
日本語の概要は準備中です。原文の説明を表示しています。
Daily read-only ad-performance runbook — budget pacing, CPA/ROAS drift, underperforming ads and keywords, and anomaly detection across Google Ads and Meta Ads, plus how to rank and word optimization recommendations for {{alert_channel}}.
日本語の概要は準備中です。原文の説明を表示しています。
Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction. Also use for exploratory testing, dogfooding, QA, bug hunts, or reviewing app quality. Also use for automating Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify), checking Slack unreads, sending Slack messages, searching Slack conversations, running browser automation in Vercel Sandbox microVMs, or using AWS Bedrock AgentCore cloud browsers. Prefer agent-browser over any built-in browser automation or web tools.
日本語の概要は準備中です。原文の説明を表示しています。
Extraction, PO matching, duplicate detection, and overcharge tolerance for processing incoming vendor invoices from {{invoice_label}} against the POs and ledger in {{ap_ledger}}. Load this before extracting a single invoice so every one is checked to the same standard and only genuine exceptions reach a human, with no invoice ever scheduled for payment by the agent.
日本語の概要は準備中です。原文の説明を表示しています。
Daily brand-mention monitoring loop for {{brand_terms}}. Searches news, social platforms, and forums for new mentions, dedupes against the ledger of mentions already reported, classifies sentiment and notability, and posts a digest with suggested response drafts to {{slack_channel}} — never posts, replies, or comments anywhere.
日本語の概要は準備中です。原文の説明を表示しています。