activity-monitoring

Real-time user activity tracking for Claude monitors environment signals and presence to adapt automation levels. It uses ManicTime data feeds and a PowerShell tool to expose a JSON context surface for decision modules.

ManicTime data feeds enable idle detection and multi-agent coordination by providing the raw activity signals needed to avoid conflicts and adapt automation timing within Windows environments.

Context-aware activity monitoring adjusts Claude's help style by learning patterns from real-time user activity. When you switch from coding to browsing, it adapts interaction timing to reduce interruptions.

The monitor-activity.ps1 tool is required to enable startup context checks in Windows environments. It processes the ManicTime data feed to populate the JSON context surface used for decision modules.

Multi-agent detection and pattern learning are scoped to Windows environments. The system requires the specific monitor-activity.ps1 tool and ManicTime data feed to function properly.

Pattern learning for context-aware decisions analyzes tracked real-time user activity to improve future interactions. It extracts context from activity switches to help Claude adapt its automation and timing over time.