proactive-claw_skill

Overview

This skill is a proactive, voice-capable life assistant that monitors your calendar, detects conflicts, and autonomously schedules prep, focus, and buffer time. It runs a user-level background daemon to scan events, send push nudges, and auto-schedule follow-ups while keeping credentials and data local. The agent includes a natural language rules and policy engine, lightweight CRM for attendees, semantic memory, and optional cross-service integrations gated by explicit flags. Setup requires Google OAuth credentials (or Nextcloud) and a one-time setup script to create a local workspace and daemon.

How this skill works

A background daemon scans your calendar on a configurable interval, runs conflict detection, and writes actionable nudges to a local pending queue. A rules and policy engine parses natural language policies and can autonomously block prep/focus/debrief windows or prompt for confirmation per policy. Local SQLite memory stores outcomes and supports TF‑IDF semantic search; optional integrations (Notion, GitHub, Telegram, LLM rater) are disabled by default and require explicit config. All credentials and tokens are stored locally under the skill workspace; outbound network calls are gated per feature flags.

When to use it

• Automatically detect and resolve calendar conflicts and back-to-back meetings
• Add prep, buffer, or debrief time around important events autonomously
• Receive push or Telegram nudges about stale action items and upcoming high-stakes meetings
• Find free time, move events, or clear safe OpenClaw calendar windows via plain English
• Run periodic energy analysis to schedule work when you historically perform best

Best practices

• Run the provided setup script once to create OAuth tokens and install the user-level daemon
• Keep feature flags disabled until you confirm each optional integration (Notion, Telegram, cross-skill, voice)
• Review policies parsed by the rules engine and set autonomous vs confirm behavior before enabling full autonomy
• Use local LLM endpoints (Ollama/LM Studio) for rating to avoid external data transfer
• Inspect setup and install scripts before running; all installed files and credentials remain in the user workspace

Example use cases

• Daemon detects overlapping events and auto-adds a 15-minute buffer before back-to-back meetings
• Before a high-stakes meeting, orchestrator gathers GitHub/Notion context and schedules a prep block
• Use natural language to find a 90-minute free slot this week and schedule focus time
• Save meeting outcomes to local SQLite and run semantic searches for past follow-ups when planning
• Enable Telegram notifications to receive mobile nudges for urgent calendar conflicts