learning-journey

name	learning-journey
description	Analyze your Copilot CLI learning journey — see how you've evolved, get personalized next-step recommendations, and estimate your ROI. Use when user asks about "my progress", "learning journey", "how am I using copilot", "what should I try next", "copilot ROI", "time saved", "copilot value", "usage evolution", "session analysis", or "copilot stats".

Learning Journey

Analyze the user's Copilot CLI session history to reveal their evolution, suggest next steps, and estimate value gained. All data comes from the local session_store database — no external services needed.

Modes

Mode	Trigger	What It Does
Full Journey	"my learning journey", "how have I evolved"	Complete analysis: timeline, milestones, patterns, recommendations, ROI
Quick Stats	"copilot stats", "usage summary"	Compact overview: session counts, patterns, current phase
Next Steps	"what should I try next", "copilot tips"	Feature recommendations based on what they haven't explored yet
ROI Estimate	"copilot ROI", "time saved", "is copilot worth it"	Value estimate with transparent assumptions

Default: If the user asks generally about their journey/progress, run Full Journey.

---

Execution Rules

1. Always query the session_store database using the sql tool with database: "session_store".
2. Never expose raw session content (summaries, messages) — only show patterns and anonymized examples.
3. Use temporal analysis — split history into time windows to show evolution, not just totals.
4. State confidence levels — be explicit when data is sparse or patterns are uncertain.
5. Frame phases as observed behavior, not absolute skill levels. Say "your usage patterns suggest..." not "you are advanced."
6. Show assumptions for any estimates. Prefer ranges over single numbers.

---

Step 1: Data Collection

Run these queries to gather the raw signals. Execute them in parallel.

Query 1: Monthly Overview

SELECT
  strftime('%Y-%m', created_at) as month,
  COUNT(*) as sessions,
  SUM((SELECT COUNT(*) FROM turns t WHERE t.session_id = s.id)) as total_turns,
  SUM((SELECT COUNT(*) FROM session_files sf WHERE sf.session_id = s.id)) as files_touched,
  ROUND(AVG((SELECT COUNT(*) FROM turns t WHERE t.session_id = s.id)), 1) as avg_turns,
  SUM((SELECT COUNT(*) FROM session_refs sr WHERE sr.session_id = s.id)) as refs_created
FROM sessions s
GROUP BY month
ORDER BY month

Query 2: Session Complexity Distribution

SELECT
  strftime('%Y-%m', s.created_at) as month,
  SUM(CASE WHEN turns <= 2 THEN 1 ELSE 0 END) as quick_sessions,
  SUM(CASE WHEN turns BETWEEN 3 AND 10 THEN 1 ELSE 0 END) as medium_sessions,
  SUM(CASE WHEN turns > 10 THEN 1 ELSE 0 END) as deep_sessions
FROM (
  SELECT s.id, s.created_at,
    (SELECT COUNT(*) FROM turns t WHERE t.session_id = s.id) as turns
  FROM sessions s
) s
GROUP BY month
ORDER BY month

Query 3: First Milestones

SELECT 'First multi-file session' as milestone,
  (SELECT MIN(s.created_at) FROM sessions s
   WHERE (SELECT COUNT(*) FROM session_files sf WHERE sf.session_id = s.id) >= 3) as achieved_at
UNION ALL
SELECT 'First commit/PR linked session',
  (SELECT MIN(s.created_at) FROM sessions s
   WHERE (SELECT COUNT(*) FROM session_refs sr WHERE sr.session_id = s.id) > 0)
UNION ALL
SELECT 'First deep session (10+ turns)',
  (SELECT MIN(s.created_at) FROM sessions s
   WHERE (SELECT COUNT(*) FROM turns t WHERE t.session_id = s.id) >= 10)
UNION ALL
SELECT 'First very deep session (50+ turns)',
  (SELECT MIN(s.created_at) FROM sessions s
   WHERE (SELECT COUNT(*) FROM turns t WHERE t.session_id = s.id) >= 50)

Query 4: Feature Detection (Advanced Usage)

SELECT
  'skills' as feature,
  COUNT(DISTINCT si.session_id) as sessions_used,
  MIN(s.created_at) as first_used
FROM search_index si
JOIN sessions s ON si.session_id = s.id
WHERE si.content LIKE '%/skills%' OR si.content LIKE '%skill%invoke%'
UNION ALL
SELECT 'MCP/plugins',
  COUNT(DISTINCT si.session_id),
  MIN(s.created_at)
FROM search_index si
JOIN sessions s ON si.session_id = s.id
WHERE si.content LIKE '%MCP%' OR si.content LIKE '%plugin%' OR si.content LIKE '%mcp-server%'
UNION ALL
SELECT 'research',
  COUNT(DISTINCT si.session_id),
  MIN(s.created_at)
FROM search_index si
JOIN sessions s ON si.session_id = s.id
WHERE si.content LIKE '%/research%'
UNION ALL
SELECT 'delegate',
  COUNT(DISTINCT si.session_id),
  MIN(s.created_at)
FROM search_index si
JOIN sessions s ON si.session_id = s.id
WHERE si.content LIKE '%/delegate%'
UNION ALL
SELECT 'code review',
  COUNT(DISTINCT si.session_id),
  MIN(s.created_at)
FROM search_index si
JOIN sessions s ON si.session_id = s.id
WHERE si.content LIKE '%/review%' OR si.content LIKE '%code-review%'
UNION ALL
SELECT 'fleet/parallel',
  COUNT(DISTINCT si.session_id),
  MIN(s.created_at)
FROM search_index si
JOIN sessions s ON si.session_id = s.id
WHERE si.content LIKE '%/fleet%' OR si.content LIKE '%parallel%subagent%'
UNION ALL
SELECT 'multi-repo',
  COUNT(DISTINCT si.session_id),
  MIN(s.created_at)
FROM search_index si
JOIN sessions s ON si.session_id = s.id
WHERE si.content LIKE '%add-dir%' OR si.content LIKE '%list-dirs%'

Query 5: Usage Consistency (Streaks)

SELECT
  DATE(created_at) as day,
  COUNT(*) as sessions_that_day
FROM sessions
GROUP BY day
ORDER BY day

Query 6: Topic Diversity

SELECT summary, created_at
FROM sessions
WHERE summary IS NOT NULL AND summary != ''
ORDER BY created_at

---

Step 2: Classify Phase (Per Time Window)

Split the user's history into thirds (early / middle / recent) by date. For each third, score these dimensions (0–3 scale):

Dimension	0 (None)	1 (Low)	2 (Medium)	3 (High)
Session depth	All ≤2 turns	Avg 3–5 turns	Avg 6–15 turns	Avg 15+ turns
File breadth	0 files/session	1–3 files avg	4–10 files avg	10+ files avg
Delivery signals	No refs	Occasional commits	Regular commits	PRs + commits
Tool diversity	Basic Q&A only	Some file ops	Skills/MCP usage	Multi-agent, plugins
Consistency	Sporadic (gaps >7d)	Weekly	Most days	Daily sustained
Topic variety	Single domain	2–3 domains	4+ domains	Cross-domain synthesis

Phase Assignment

Total Score	Phase	Label
0–5	🌱 Explorer	"Discovering what's possible"
6–10	🔨 Builder	"Actively building with AI assistance"
11–15	🎯 Orchestrator	"Directing AI as a force multiplier"
16–18	🚀 Architect	"Designing systems of AI-assisted workflows"

Important: Show the user's phase trajectory across the three time windows, not just their current state.

---

Step 3: Identify Milestones

Present milestones as a timeline. Good milestone types:

• 📅 First session ever
• 📁 First multi-file change
• 🔗 First commit/PR linked session
• 🏗️ First deep session (10+ turns)
• 🧩 First skill/plugin usage
• 🔄 First sustained streak (3+ consecutive days)
• 🌐 First cross-domain usage (different repos/topics in same week)
• 🤖 First multi-agent / delegation usage

Format as a simple timeline:

📅 Feb 24 — First session (Agent Framework exploration)
📁 Feb 24 — First multi-file build (46 files in one session!)
🏗️ Feb 27 — First marathon session (44 turns)
...

---

Step 4: Next Steps Recommendations

Based on Query 4 (feature detection), identify features not observed in their history. Recommend the top 2–3 most impactful ones based on their current phase:

Current Phase	Best Next Features
🌱 Explorer	Multi-file editing, using @file mentions, iterating on code
🔨 Builder	/review for code quality, skills for domain tasks, git integration
🎯 Orchestrator	/delegate to create PRs, /fleet for parallel work, custom MCP servers
🚀 Architect	Building custom skills, sharing plugins, teaching others

For each recommendation:

1. Name the feature
2. Explain what it does in one sentence
3. Give a concrete example scenario relevant to their observed usage patterns
4. Estimate the unlock potential ("This could save you ~X min per session where you do Y")

Important: Say "not observed in your session history" — NOT "you haven't used this." They may use it outside tracked sessions.

---

Step 5: ROI Estimate

Methodology (Be Transparent)

Present ROI as an illustrative estimate with clear assumptions. Use this structure:

Time Saved Estimation

Classify each session by dominant type and apply the appropriate multiplier:

Session Type	Criteria	Conservative	Moderate	Aggressive
Quick Q&A	≤2 turns, 0 files	2 min saved	5 min saved	8 min saved
Code generation	1–5 files touched	10 min saved	20 min saved	35 min saved
Deep build	5+ files, 10+ turns	30 min saved	60 min saved	120 min saved
Workflow/automation	refs, skills, plugins	20 min saved	45 min saved	90 min saved

Do NOT stack multipliers. Each session counts once under its dominant type.

Cost Context

Plan	Monthly Cost	Analysis Period Cost
Individual	$19/month	Calculate for their time span
Business	$39/month	Calculate for their time span

Output Format

📊 ROI Estimate (your {N} months of usage)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Sessions analyzed: {total}
  Quick Q&A: {n} sessions
  Code generation: {n} sessions
  Deep builds: {n} sessions
  Workflow/automation: {n} sessions

⏱️ Estimated time saved:
  Conservative: {X} hours ({Y} hours/month)
  Moderate:     {X} hours ({Y} hours/month)
  Aggressive:   {X} hours ({Y} hours/month)

💰 At your hourly rate:
  (User can multiply by their rate — we don't assume income)

📈 Cost comparison (Individual plan):
  Tool cost: ${cost} over {months} months
  Break-even: ~{N} min/day of saved time
  Your moderate estimate: ~{M} min/day

⚠️ Assumptions: See methodology above. Actual value depends on
   task complexity, domain expertise, and alternative approaches.

---

Step 6: Compose Output

Full Journey Format

🗺️ Your Copilot CLI Learning Journey
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📊 Overview
  {months} months • {sessions} sessions • {files} files touched • {refs} deliverables

🌱→🔨→🎯 Your Evolution
  [Phase trajectory with scores per time window]

🏆 Key Milestones
  [Timeline of firsts]

📈 Patterns
  [2-3 interesting observations about their usage]

💡 Recommended Next Steps
  [Top 2-3 feature recommendations]

💰 ROI Estimate
  [Structured estimate with ranges]

Sparse Data Mode

If the user has fewer than 5 sessions, use a simplified output:

🗺️ Your Copilot CLI Snapshot
━━━━━━━━━━━━━━━━━━━━━━━━━━━

You're just getting started! Here's what I see so far:
  {sessions} sessions over {days} days

🌱 Current Phase: Explorer
  [Brief description of what they've done]

💡 Quick Wins to Try Next
  [3 accessible features for beginners]

📊 Come back after 10+ sessions for a full journey analysis!

---

Presentation Rules

1. Use emoji sparingly for section headers — don't overload.
2. Keep the tone encouraging and forward-looking, never judgmental.
3. Celebrate progress — highlight what's impressive about their evolution.
4. Be specific about patterns ("You went from 2 sessions/week to daily usage in March").
5. Never show raw session content, messages, or sensitive project names.
6. If topic summaries contain personal/work info, generalize them ("professional communication", "web development project").
7. Round numbers — "~50 sessions" not "47 sessions."
8. Always end with an actionable next step.