Skills Library View on GitHub

Other

/nightly-journal

End-of-day client/work journal. Triggers: nightly journal, daily sweep, check comms, what happened today.

$ golems-cli skills install nightly-journal

Updated 2 weeks ago

Autonomous sweep of WhatsApp (Business + Personal), Gmail, and BrainLayer for professional activity. Writes a structured daily log to Obsidian.

Scope

Include: Client chats, job pipeline (recruiters, interviews, applications), networking (TechGym, LinkedIn), outreach responses.

Exclude: Personal chats, group spam, newsletters, GitHub notifications, marketing, status broadcasts.

Sources

WhatsApp Business — client JIDs from BrainLayer (brain_search("whatsapp jids known contacts")). Query each with mcp__whatsapp__list_messages(chat_jid=JID, limit=30). Also scan mcp__whatsapp__list_chats(limit=20, sort_by="last_active") for new contacts.
WhatsApp Personal — project JIDs from BrainLayer. Same query pattern via mcp__whatsapp__list_messages.
Gmail — gmail_search_messages(q="after:<today> -category:promotions -category:social"). Keep only recruiter/client/interview emails.
BrainLayer — brain_search("owner-state career", date_from=today) for pipeline context.

Output

Write to: ~/Library/Mobile Documents/iCloud~md~obsidian/Documents/personal/Personal/<YYYY-MM-DD> Daily Log.md

If file exists, append ## Evening Sweep — don't overwrite.

Sections: Clients, Job Pipeline (active conversations + applications + interviews), Networking, Email Highlights, Tomorrow (what needs attention first). Skip empty sections.

Summarize Hebrew content in English (RTL breaks some editors). End with brain_store tagged ["daily-journal", "agent:coachClaude"].

Full SKILL.md source — includes LLM directives, anti-patterns, and technical instructions stripped from the Overview tab.

Autonomous sweep of WhatsApp (Business + Personal), Gmail, and BrainLayer for professional activity. Writes a structured daily log to Obsidian.

Scope

Include: Client chats, job pipeline (recruiters, interviews, applications), networking (TechGym, LinkedIn), outreach responses.

Exclude: Personal chats, group spam, newsletters, GitHub notifications, marketing, status broadcasts.

Sources

WhatsApp Business — client JIDs from BrainLayer (brain_search("whatsapp jids known contacts")). Query each with mcp__whatsapp__list_messages(chat_jid=JID, limit=30). Also scan mcp__whatsapp__list_chats(limit=20, sort_by="last_active") for new contacts.
WhatsApp Personal — project JIDs from BrainLayer. Same query pattern via mcp__whatsapp__list_messages.
Gmail — gmail_search_messages(q="after:<today> -category:promotions -category:social"). Keep only recruiter/client/interview emails.
BrainLayer — brain_search("owner-state career", date_from=today) for pipeline context.

Output

Write to: ~/Library/Mobile Documents/iCloud~md~obsidian/Documents/personal/Personal/<YYYY-MM-DD> Daily Log.md

If file exists, append ## Evening Sweep — don't overwrite.

Sections: Clients, Job Pipeline (active conversations + applications + interviews), Networking, Email Highlights, Tomorrow (what needs attention first). Skip empty sections.

Summarize Hebrew content in English (RTL breaks some editors). End with brain_store tagged ["daily-journal", "agent:coachClaude"].

Data Integrity

WhatsApp MCP returns empty for valid JIDs — known issue. Log "MCP sync issue, check phone" rather than "no activity."
Parallel queries — run all WhatsApp + Gmail queries in parallel (independent data sources).
Cross-validate — if BrainLayer has an owner-state connection entry but WhatsApp shows no messages, note the gap. The absence of data is data.
Never fabricate — if a source didn't return data, say so. Don't infer activity from BrainLayer alone.

Urgency Detection

If an unanswered client ask is found (message from client with no reply from Etan), flag with ⚠️ in the journal output and mention it prominently at the top.

This skill is read-only — it reads sources and writes to Obsidian journal only. It never sends WhatsApp messages, emails, or takes external actions.