Weekly Report

Weekly Support Report

Generate a complete support operations snapshot by combining data from multiple analytics tools.

Workflow Modes

Current Period (default)

Generates a full report for a single time period.

1. Determine the period from the user's request. Default: this_week. Supported values: this_week, last_week, this_month, last_month, last_7_days, last_30_days.
2. Fetch dashboard KPIs using get_metrics with the resolved period
3. Fetch team performance using get_agent_performance with the same period
4. Fetch channel breakdown using get_channel_analytics with the same period
5. Fetch conversation trends using get_trends with metric: "conversations" and the period
6. Fetch AI automation trends using get_trends with metric: "ai_responses" and the period
7. Compile and present the full report

Comparison Mode

Compares two periods side by side with delta calculations showing improvement or regression.

1. Determine the two periods (e.g., this_week vs last_week, or this_month vs last_month)
2. Fetch all data for both periods (KPIs, team performance, channels, trends)
3. Calculate deltas for each metric: absolute change and percentage change
4. Flag metrics that improved with an up arrow and metrics that regressed with a down arrow
5. Present the comparison table with both periods and the delta column

Executive Summary

A condensed version for leadership — key numbers, top-line trends, and action items only. No detailed tables.

1. Fetch dashboard KPIs using get_metrics with the resolved period
2. Fetch team performance using get_agent_performance (top 3 performers only)
3. Fetch conversation trends using get_trends with metric: "conversations"
4. Present a 5-8 line summary with: total volume, resolution rate, average response time, AI automation rate, top performer, and one key insight or action item

Report Structure

Support Operations — Weekly Report

Overview ([Period])

• Open conversations: [count]
• Resolved this period: [count]
• New contacts: [count]
• Total messages: [count]
• AI-handled responses: [count] ([percentage]% automation rate)
• Avg first response time: [duration]
• Avg resolution time: [duration]

Team Performance

Channel Breakdown

Trends

• Conversation volume trend: [increasing/stable/decreasing]
• AI automation trend: [improving/stable/declining]
• Response time trend: [improving/stable/declining]

Key Insights

• Notable observations: spikes, top performers, channels needing attention
• Recommended actions based on the data

Comparison Report Additions

When in Comparison Mode, add a delta column to each table:

Examples

• "Weekly report" — Runs Current Period mode for this_week. Produces the full report.
• "Compare this week to last week" — Runs Comparison Mode. Shows both periods side by side with deltas.
• "Executive summary for this month" — Runs Executive Summary mode for this_month. Produces a concise 5-8 line summary.
• "How did we do last week?" — Runs Current Period mode for last_week.
• "Monthly report with comparison to last month" — Runs Comparison Mode for this_month vs last_month.

Edge Cases

• No data for the requested period: Report "No data available for [period]" and suggest trying a different time range or checking that the Corebee integration is active.
• Partial data (some API calls succeed, others fail): Present the data that is available and clearly note which sections are missing. Do not fabricate numbers.
• Missing or inactive agents: If get_agent_performance returns agents with zero activity, include them in the table but note them as inactive. Do not exclude them silently.
• New team (first week): If there is no prior period for comparison, skip Comparison Mode and note that baseline data is being established.
• Very high volume (1000+ conversations): Summarize by category rather than listing individual conversations. Focus on aggregates and percentages.
• API errors: Report the specific error and suggest retrying. If get_metrics fails, the report cannot be generated — inform the user.

Settings

Cross-Skill References

• Use /metrics-compare for deeper drill-down analysis on specific metrics across custom date ranges.
• Use /inbox-triage after reviewing the report to act on open conversations that need attention.

Guidelines

• Always use agent names from profile data, not IDs (unless show_ids is enabled)
• Calculate AI automation rate as: ai_responses / total_messages * 100
• Keep the tone professional but conversational
• Highlight both wins (high resolution rates, fast response times) and areas for improvement
• In Comparison Mode, use clear directional indicators for deltas (up/down arrows or +/- signs)
• Respect the configured timezone when determining period boundaries
• When generating an Executive Summary, prioritize actionable insights over raw numbers
• Fetch all data sources in parallel where possible to minimize latency