ワンクリックで
gmail-search-context
Search Gmail intelligently — construct precise queries, resolve threads, handle attachments, synthesize findings
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Search Gmail intelligently — construct precise queries, resolve threads, handle attachments, synthesize findings
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Analyze spreadsheet data intelligently - understand structure, choose right analysis tool (SQL/pivot/chart), apply formatting, present insights.
Create charts and visualizations in Google Sheets — detect data structure, select chart type, format for clarity, present insights
Generate an editable Microsoft Word (.docx) document — reports, letters, memos, structured docs. Use when the user wants a Word file. Built with docx-js (Node) for high fidelity.
Generate a polished, printable PDF — reports, letters, invoices, resumes, one-pagers. Use when the user wants a PDF document. Typst is the primary engine; LaTeX (tectonic) is the fallback for niche packages or specific academic/journal templates.
Generate a PowerPoint (.pptx) slide deck — pitch decks, reviews, summaries. Use when the user wants slides. Built with pptxgenjs (Node) for native, editable slides and charts.
Generate an Excel (.xlsx) workbook or a CSV file — tables, financial models, data exports, formatted reports with charts. Use when the user wants a spreadsheet or CSV. Built with openpyxl + pandas (Python).
| name | gmail-search-context |
| description | Search Gmail intelligently — construct precise queries, resolve threads, handle attachments, synthesize findings |
| target | gmail_agent |
User wants to find specific emails, search for information in their inbox, or gather email context on a topic.
Gmail supports powerful search operators:
People:
from:user@example.com — emails fromto:user@example.com — emails tocc:user@example.com — CC'd toStatus:
is:unread / is:read / is:starred / is:important / is:snoozedCategories:
category:primary / category:social / category:promotions / category:updateslabel:custom-label — user-created labels onlyTime:
after:2025/01/01 / before:2025/02/01newer_than:7d / older_than:30dafter:YYYY/MM/DD before:YYYY/MM/DD+1 (e.g., all emails on Jan 15: after:2025/01/15 before:2025/01/16)Content:
subject:meeting — in subject line"exact phrase" — exact matchhas:attachment / filename:pdf / filename:xlsxlarger:5M / smaller:1M — by sizeLogic: AND (default), OR, -exclude
Always use spawn_subagent for any Gmail fetch — raw email responses are too large for the parent context regardless of result count. The subagent fetches, summarizes, and returns only a compact digest + next_page_token.
Find specific emails:
result = spawn_subagent(
task="""
Call GMAIL_FETCH_EMAILS(query="from:sarah@company.com subject:Q1 after:2025/01/01", max_results=30)
Summarize all emails: sender, subject, date, key points, action items.
digest: <summary>, next_page_token: <token or null>
"""
)
Find contacts (lightweight — safe to call directly):
GMAIL_SEARCH_PEOPLE(query="Sarah", pageSize=10)
Thread-based search:
result = spawn_subagent(
task="""
Call GMAIL_LIST_THREADS(query="project proposal from:alex", max_results=30, verbose=true)
For each thread extract: participants, timeline, key decisions, action items.
digest: <summary>, next_page_token: <token or null>
"""
)
Fetch ALL emails for a specific day:
Use after:YYYY/MM/DD before:YYYY/MM/DD+1. Parent orchestrates pagination — spawn a subagent per page, each returning a digest + token. Parent spawns the next only when a token is returned:
# Page 1
result_1 = spawn_subagent(
task="""
Call GMAIL_FETCH_EMAILS(query="after:2025/01/15 before:2025/01/16", max_results=30)
Summarize all emails: sender, subject, date, key points, action items.
digest: <summary>, next_page_token: <token or null>
"""
)
# Page 2 — only if token returned
if result_1.next_page_token:
result_2 = spawn_subagent(
task="""
Call GMAIL_FETCH_EMAILS(query="after:2025/01/15 before:2025/01/16", max_results=30, page_token="<token>")
Same summarization instructions.
digest: <summary>, next_page_token: <token or null>
"""
)
# Repeat until next_page_token is null. Parent synthesizes all digests.
"quarterly report from:finance@company.com after:2025/01/01 has:attachment""quarterly report from:finance@company.com""quarterly report has:attachment""quarterly report"Parent collects all subagent digests and presents organized results:
Found 8 emails about "Q1 budget proposal":
Thread: "Q1 Budget Review" (5 messages)
From: Sarah → Finance Team | Jan 15-22
Summary: Initial proposal → revision → final approval
Attachment: Q1_Budget_Final.xlsx (in latest message)
Status: Approved (Sarah's last message: "Looks good, approved.")
Thread: "Budget Follow-up" (3 messages)
From: Alex → Sarah, You | Jan 25
Summary: Questions about marketing allocation
Status: Awaiting your response
GMAIL_FETCH_EMAILS or GMAIL_LIST_THREADS in the parent context — always use spawn_subagentlabel:snoozed (wrong — use is:snoozed)