Dev Flow Analytics

Track and analyze how the Dev Flow system is being used. Provides metrics on skill invocations, agent performance, and thread outcomes.

When to Use

•User asks about skill usage patterns
•User wants to see performance metrics
•Generating reports on Dev Flow effectiveness
•Identifying underutilized skills
•Tracking thread success rates
•Measuring productivity improvements

Input Handling

For Reports:

•period: Time period (today, week, month, all)
•focus: Specific area (skills, agents, threads, all)
•format: Output format (summary, detailed, json)

For Tracking (automatic):

•Skill invocations are logged automatically
•Thread outcomes are captured on completion
•Agent delegations are recorded

Instructions

Data Collection

Analytics data is stored in ~/.claude/analytics/:

code

~/.claude/analytics/
├── skill-usage.jsonl      # Skill invocation log
├── agent-metrics.json     # Agent performance summary
├── thread-history.jsonl   # Thread outcomes
└── daily-summary.json     # Rolling daily stats

Skill Usage Schema (skill-usage.jsonl)

json

{
  "timestamp": "2026-02-06T10:30:00Z",
  "skill": "launch-planner",
  "trigger": "build new feature",
  "duration_ms": 4500,
  "outcome": "success",
  "context": "Capsule Explore",
  "thread_id": "abc123"
}

Agent Metrics Schema (agent-metrics.json)

json

{
  "agents": {
    "implementer": {
      "tasks_completed": 45,
      "avg_duration_ms": 12000,
      "success_rate": 0.93,
      "common_tasks": ["code changes", "file edits", "refactoring"]
    },
    "code-reviewer": {
      "reviews_completed": 38,
      "ship_rate": 0.84,
      "avg_issues_found": 2.3
    }
  },
  "updated": "2026-02-06T10:30:00Z"
}

Thread History Schema (thread-history.jsonl)

json

{
  "thread_id": "abc123",
  "type": "b",
  "goal": "Add dark mode toggle",
  "started": "2026-02-06T10:00:00Z",
  "completed": "2026-02-06T10:45:00Z",
  "outcome": "shipped",
  "validation_iterations": 2,
  "skills_fired": ["launch-planner", "design-guide", "test-engineer"],
  "agents_used": ["implementer", "code-reviewer", "validator"]
}

Report Generation

Summary Report (Default)

markdown

## Dev Flow Analytics Report

**Period:** Last 7 days
**Generated:** 2026-02-06

### Overview

| Metric | Value | Trend |
|--------|-------|-------|
| Total Skill Invocations | 156 | ↑ 12% |
| Unique Skills Used | 24 | → 0% |
| Thread Success Rate | 89% | ↑ 5% |
| Avg Thread Duration | 18 min | ↓ 8% |

### Top Skills

| Rank | Skill | Invocations | Success Rate |
|------|-------|-------------|--------------|
| 1 | launch-planner | 34 | 100% |
| 2 | test-engineer | 28 | 96% |
| 3 | code-reviewer | 25 | 100% |
| 4 | design-guide | 18 | 94% |
| 5 | implementer | 15 | 93% |

### Underutilized Skills

These skills haven't been used recently:
- `competitive-intelligence` (14 days)
- `revenue-optimizer` (10 days)
- `accessibility-auditor` (7 days)

### Agent Performance

| Agent | Tasks | Success | Avg Duration |
|-------|-------|---------|--------------|
| implementer | 45 | 93% | 12s |
| code-reviewer | 38 | 100% | 8s |
| validator | 52 | 87% | 5s |

### Thread Outcomes

| Type | Count | Shipped | Avg Iterations |
|------|-------|---------|----------------|
| base | 28 | 89% | 1.2 |
| b (big) | 12 | 83% | 2.4 |
| p (parallel) | 5 | 100% | 1.0 |

### Recommendations

1. **Try `competitive-intelligence`** - Useful for market analysis
2. **Run accessibility audit** - Last check was 7 days ago
3. **Thread validation improving** - Down from 2.8 to 2.4 iterations

Detailed Report

Includes:

•Full skill invocation log
•Individual thread breakdowns
•Time-series charts (ASCII)
•Skill chain effectiveness

Tracking Functions

Log Skill Invocation:

javascript

// Called automatically by organism
logSkillInvocation({
  skill: "launch-planner",
  trigger: userMessage,
  startTime: Date.now()
});

Log Thread Outcome:

javascript

// Called on thread completion
logThreadOutcome({
  threadId: currentThread.id,
  type: currentThread.type,
  outcome: "shipped", // or "abandoned", "blocked"
  iterations: validationCount
});

Update Agent Metrics:

javascript

// Called after agent task
updateAgentMetrics({
  agent: "implementer",
  task: "code changes",
  duration: elapsed,
  success: true
});

Required Output Sections

Analytics Summary

markdown

## Dev Flow Analytics

**Period:** {{period}}
**Skills Used:** {{unique_skills}} / {{total_skills}}
**Success Rate:** {{success_rate}}%

### Quick Stats
- Most used: {{top_skill}}
- Least used: {{bottom_skill}}
- Busiest day: {{busiest_day}}
- Total threads: {{thread_count}}

Recommendations

markdown

## Recommendations

Based on your usage patterns:

1. **{{recommendation_1}}**
   - Why: {{reason_1}}
   - Action: {{action_1}}

2. **{{recommendation_2}}**
   - Why: {{reason_2}}
   - Action: {{action_2}}

Guardrails

•Never log sensitive data (secrets, passwords, PII)
•Aggregate metrics, don't store raw conversation content
•Respect user privacy - analytics are local only
•Allow users to clear analytics data
•Don't track during "quick:" prefixed requests

Commands

/analytics

code

/analytics                    # Summary for last 7 days
/analytics today              # Today's stats
/analytics month detailed     # Full month breakdown
/analytics skills             # Skill-focused report
/analytics threads            # Thread-focused report
/analytics clear              # Clear all analytics data

Trigger Patterns

javascript

if (/(analytics|usage stats|skill metrics|performance report|dev flow stats|how am I using|\\/analytics)/i.test(objective)) {
  // Run devflow-analytics
}

Integration

Dependencies: None (standalone tracking) Triggers: Runs passively, reports on demand Data Location: ~/.claude/analytics/

Version History

•v1.0.0: Initial release with skill, agent, and thread tracking