Vibecosystem session-analysis-patterns
Claude Code session analysis, tool call patterns, efficiency metrics, and anti-pattern detection
install
source · Clone the upstream repo
git clone https://github.com/vibeeval/vibecosystem
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/vibeeval/vibecosystem "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/session-analysis-patterns" ~/.claude/skills/vibeeval-vibecosystem-session-analysis-patterns && rm -rf "$T"
manifest:
skills/session-analysis-patterns/SKILL.mdsource content
Session Analysis Patterns
Session Metrics
| Metrik | İyi | Orta | Kötü |
|---|---|---|---|
| Tokens per task | <50K | 50-100K | >100K |
| Tool calls per task | <20 | 20-40 | >40 |
| Repeated reads | 0-1 | 2-3 | >3 |
| Blind edits | 0 | 1 | >1 |
| Agent spawn efficiency | >80% useful | 50-80% | <50% |
Anti-Pattern Catalog
| # | Pattern | Tespit | Fix |
|---|---|---|---|
| 1 | Repeated Read | Aynı dosya 3+ kez okunmuş | Bilgiyi ilk seferde kaydet |
| 2 | Wide Grep | Pattern'siz geniş arama | Spesifik pattern + glob filter |
| 3 | Blind Edit | Dosya okumadan edit | Read → Edit sırası zorunlu |
| 4 | Idle Agent | Agent spawn edilip sonuç kullanılmamış | Gereksiz spawn etme |
| 5 | Cascade Fail | Aynı hata 3+ kez retry | Root cause analiz et |
| 6 | Context Pollution | Ana context'te gereksiz büyük dosya okuma | Agent'a delege et |
| 7 | Premature Agent | Basit iş için agent spawn | Direkt tool kullan |
| 8 | Missing Parallel | Bağımsız işler sıralı | Paralel agent/tool call |
| 9 | Over-Engineering | Basit fix için büyük refactoring | Minimal değişiklik |
Efficiency Scorer (0-100)
def calculate_efficiency(session): score = 100 # Penalties score -= session.repeated_reads * 5 # -5 per repeated read score -= session.blind_edits * 15 # -15 per blind edit score -= session.wide_greps * 3 # -3 per wide grep score -= session.idle_agents * 10 # -10 per idle agent score -= session.cascade_fails * 8 # -8 per cascade fail # Bonuses score += session.parallel_calls * 2 # +2 per parallel batch score += session.first_pass_success * 5 # +5 per first-try success return max(0, min(100, score))
Session Comparison
| Metrik | Session A | Session B | Delta | |--------|-----------|-----------|-------| | Total tokens | 45,000 | 120,000 | +167% | | Tool calls | 12 | 38 | +217% | | Time (min) | 3 | 12 | +300% | | Tasks completed | 2 | 2 | 0% | | Efficiency score | 88 | 42 | -52% |
Braintrust Log Analysis
# Session analizi btca sessions list --last 7d btca session <id> --metrics # Pattern detection btca session <id> --anti-patterns # Cost analysis btca sessions list --sort cost --desc
Checklist
- Session token usage tracked
- Anti-pattern detection aktif
- Efficiency score > 70
- Repeated reads < 2
- Blind edits = 0
- Parallel opportunities exploited
- Agent spawn ROI positive
Anti-Patterns
- Her session'ı analiz etmek (sadece outlier'ları analiz et)
- Metric'lere takılıp task quality'yi ihmal etmek
- Hız için correctness'tan ödün vermek