activity-patterns

Activity Patterns

Analyze the user's Claude Code usage patterns over time to provide personalized scheduling insights.

Data Source

Read ${CLAUDE_PLUGIN_DATA}/activity_patterns.json — an array of session start records:

[
  {"date": "2026-03-26", "time": "22:15", "weekday": "Wednesday", "hour": 22},
  {"date": "2026-03-27", "time": "08:10", "weekday": "Thursday", "hour": 8}
]

The SessionStart hook appends an entry each time a new session begins. Max 200 entries retained.

Analysis to Perform

Time Distribution

• Count sessions by hour of day (0-23)
• Find the peak activity hours
• Identify dead hours (when they never work)

Day Distribution

• Sessions per weekday
• Weekday vs weekend ratio
• Most active day of the week

Pattern Classification

Based on the data, classify the user:

Pattern	Criteria
Night Owl	>50% of sessions between 8pm-4am
Early Bird	>50% of sessions between 5am-10am
Business Hours	>50% of sessions between 9am-5pm
All-Hours	No dominant time block
Weekend Warrior	>30% of sessions on weekends

Peak Hour Overlap

• What percentage of their sessions fall during peak hours?
• Are they naturally avoiding peak, or hitting it head-on?
• Personalized recommendation based on their actual pattern

Response Format

Activity Profile: [N] sessions analyzed over [date range]

Schedule: [Night Owl / Early Bird / Business Hours / All-Hours]
Most active: [day of week] at [hour range]
Least active: [day/hour]
Peak hour overlap: [X]% of your sessions hit peak hours

[Insight]: [Personalized observation, e.g., "You naturally work off-peak —
your night schedule dodges throttling completely."]

[Recommendation]: [Actionable advice based on pattern]

Privacy Note

Activity patterns are stored locally in ${CLAUDE_PLUGIN_DATA}. Nothing is sent anywhere. The data stays on the user's machine and is only used to provide personalized scheduling insights.