contact-deep-dive

name	contact-deep-dive
description	Comprehensive analysis of your messaging relationship with a specific person — frequency, patterns, reactions, shared content, and conversation highlights.

Contact Deep Dive

Build a full picture of your messaging relationship with one person.

When to Use

When the user asks about a specific person:

• "Tell me about my texting relationship with Alice"
• "How much do I talk to [person]?"
• "Give me stats on [contact]"
• "What's my messaging history with [name]?"

Steps

1 — Identify the contact

search_contacts({ query: "<name the user provided>" })

If multiple matches, ask the user to confirm which one. Use the exact name value returned.

2 — Pull all stats in parallel

Run these tools together with contact set to the resolved name:

• contact_stats({ contact: NAME }) — totals, first/last date, active days, avg per day
• time_series({ contact: NAME, window: 28 }) — trend over time
• reactions({ contact: NAME }) — reactions exchanged
• top_contacts({ limit: 1, direct_only: true }) — confirm rank among all contacts

Also run:

-- Sent vs received breakdown
SELECT
  SUM(CASE WHEN is_from_me = 1 THEN 1 ELSE 0 END) AS sent,
  SUM(CASE WHEN is_from_me = 0 THEN 1 ELSE 0 END) AS received,
  ROUND(100.0 * SUM(CASE WHEN is_from_me = 1 THEN 1 ELSE 0 END) / COUNT(*), 1) AS pct_sent
FROM messages WHERE name = '<NAME>'

-- Most active years
SELECT year, COUNT(*) AS n FROM messages
WHERE name = '<NAME>'
GROUP BY year ORDER BY year

-- Day of week pattern
SELECT
  CASE EXTRACT(DOW FROM CAST(timestamp AS TIMESTAMP))
    WHEN 0 THEN 'Sunday' WHEN 1 THEN 'Monday' WHEN 2 THEN 'Tuesday'
    WHEN 3 THEN 'Wednesday' WHEN 4 THEN 'Thursday' WHEN 5 THEN 'Friday'
    WHEN 6 THEN 'Saturday'
  END AS day,
  COUNT(*) AS n
FROM messages WHERE name = '<NAME>' AND timestamp IS NOT NULL
GROUP BY EXTRACT(DOW FROM CAST(timestamp AS TIMESTAMP)), day
ORDER BY EXTRACT(DOW FROM CAST(timestamp AS TIMESTAMP))

3 — Synthesise into a narrative

Present the results as a coherent story, not a list of numbers. Structure:

Relationship overview — when the relationship started, total volume, how they rank among your contacts, whether you text more or they text more.

Trajectory — is messaging increasing, decreasing, or steady? Call out any notable spikes or gaps.

Patterns — most active day of week, whether it's a morning/evening relationship (if visible from data).

Reactions — what reactions you exchange most; who sends more Loved vs Liked etc.

One interesting finding — surface the most surprising or meaningful data point.

End by offering to drill into a specific year or compare to the overall average.