Agent hata toleransi ve devre kesici pattern. Ust uste hata yapan agent'i durdur, cooldown uygula, fallback'e gec. Kaskatli hatalari ve sonsuz retry dongularini onler.
npx skills add https://github.com/vibeeval/vibecosystem --skill circuit-breaker
Agent'lar da servisler gibi basarisiz olabilir. Ayni hatayi tekrar tekrar denemek token israf eder ve sorunu cozmez. Circuit breaker bunu onler.
CLOSED (Normal)
Agent calisir, hatalar sayilir.
Hata esigi asilirsa → OPEN'a gec.
OPEN (Devre Kesik)
Agent CALISTIRILMAZ.
Cooldown suresi boyunca bekle.
Cooldown bitince → HALF-OPEN'a gec.
HALF-OPEN (Test)
Tek bir istek gonder.
Basarili → CLOSED'a don.
Basarisiz → OPEN'a geri don (cooldown uzat).
basarili hata esigi
┌──────────┐ ┌───────────┐
│ │ │ │
▼ │ ▼ │
CLOSED ──────┼── OPEN ────── HALF-OPEN
▲ │ │ │
│ │ │ │
└──────────┘ └───────────┘
normal cooldown bitti
interface CircuitBreakerConfig {
failureThreshold: number // Kac hata sonrasi OPEN (default: 3)
cooldownMs: number // OPEN'da bekleme suresi (default: 60000 = 1 dk)
halfOpenMaxAttempts: number // HALF-OPEN'da max deneme (default: 1)
resetAfterMs: number // Hata sayacini sifirla (default: 300000 = 5 dk)
onOpen?: () => void // OPEN'a gecince cagrilir
onClose?: () => void // CLOSED'a donunce cagrilir
}
const DEFAULT_CONFIG: CircuitBreakerConfig = {
failureThreshold: 3,
cooldownMs: 60000,
halfOpenMaxAttempts: 1,
resetAfterMs: 300000,
}
class AgentCircuitBreaker {
private state: 'CLOSED' | 'OPEN' | 'HALF-OPEN' = 'CLOSED'
private failures = 0
private lastFailureTime = 0
private config: CircuitBreakerConfig
constructor(private agentName: string, config?: Partial<CircuitBreakerConfig>) {
this.config = { ...DEFAULT_CONFIG, ...config }
}
canExecute(): boolean {
if (this.state === 'CLOSED') return true
if (this.state === 'OPEN') {
const elapsed = Date.now() - this.lastFailureTime
if (elapsed >= this.config.cooldownMs) {
this.state = 'HALF-OPEN'
return true
}
return false
}
// HALF-OPEN: tek denemeye izin ver
return true
}
recordSuccess(): void {
this.failures = 0
if (this.state === 'HALF-OPEN') {
this.state = 'CLOSED'
this.config.onClose?.()
}
}
recordFailure(): void {
this.failures++
this.lastFailureTime = Date.now()
if (this.state === 'HALF-OPEN') {
this.state = 'OPEN'
return
}
if (this.failures >= this.config.failureThreshold) {
this.state = 'OPEN'
this.config.onOpen?.()
}
}
getStatus(): { state: string; failures: number; agent: string } {
return { state: this.state, failures: this.failures, agent: this.agentName }
}
}
const breakers: Record<string, AgentCircuitBreaker> = {
'code-reviewer': new AgentCircuitBreaker('code-reviewer', { failureThreshold: 3 }),
'security-reviewer': new AgentCircuitBreaker('security-reviewer', { failureThreshold: 2 }),
'sleuth': new AgentCircuitBreaker('sleuth', { failureThreshold: 3 }),
}
async function spawnAgent(name: string, task: string): Promise<string> {
const breaker = breakers[name]
if (!breaker?.canExecute()) {
console.warn(`Circuit OPEN: ${name} -- fallback kullaniliyor`)
return executeFallback(name, task)
}
try {
const result = await executeAgent(name, task)
breaker.recordSuccess()
return result
} catch (error) {
breaker.recordFailure()
console.error(`${name} basarisiz (${breaker.getStatus().failures}/${3})`)
if (!breaker.canExecute()) {
return executeFallback(name, task)
}
throw error
}
}
Agent devre disiyken ne yapilacagi:
| Agent | Fallback 1 | Fallback 2 | Fallback 3 |
|-------|-----------|-----------|-----------|
| code-reviewer | Manuel Grep review | Basit lint calistir | Kullaniciya bildir |
| security-reviewer | Grep ile secret scan | SAST tool calistir | Kullaniciya bildir |
| sleuth | scout ile arastir | Manuel debug | Kullaniciya bildir |
| kraken | spark ile parcali fix | Manuel implement | Kullaniciya bildir |
| verifier | Manuel build + test | Sadece build kontrol | Kullaniciya bildir |
| architect | planner ile basit plan | Kullaniciya sor | - |
| build-error-resolver | Manuel hata oku + fix | Kullaniciya bildir | - |
Her hata ayni agirlikta degil:
| Hata Tipi | Sayac Etkisi | Ornek |
|-----------|:------------:|-------|
| API timeout | +1 | Anthropic API timeout |
| Rate limit | +0 (beklenir) | 429 Too Many Requests |
| Invalid output | +1 | Agent bos cikti verdi |
| Tool error | +0.5 | Bash komutu basarisiz |
| Logic error | +2 | Agent yanlis dosyayi duzenledi |
| Crash | +3 | Agent tamamen cokktu |
#!/bin/bash
# scripts/circuit-status.sh
echo "=== Agent Circuit Breaker Status ==="
echo ""
printf "%-25s %-10s %-10s\n" "Agent" "State" "Failures"
echo "-------------------------------------------"
# Canavar skill-matrix'ten oku
if [ -f ~/.claude/canavar/skill-matrix.json ]; then
cat ~/.claude/canavar/skill-matrix.json | \
jq -r '.agents | to_entries[] | "\(.key) \(.value.failures // 0) \(.value.state // "CLOSED")"' | \
while read name failures state; do
if [ "$state" = "OPEN" ]; then
printf "%-25s \033[31m%-10s\033[0m %-10s\n" "$name" "$state" "$failures"
elif [ "$state" = "HALF-OPEN" ]; then
printf "%-25s \033[33m%-10s\033[0m %-10s\n" "$name" "$state" "$failures"
else
printf "%-25s \033[32m%-10s\033[0m %-10s\n" "$name" "CLOSED" "$failures"
fi
done
fi
WARN: Agent 2+ ust uste basarisiz
ERROR: Circuit OPEN'a gecti
CRIT: 3+ agent ayni anda OPEN (sistemik sorun)
Tekrarlayan hatalarda cooldown suresini artir:
1. hata → 1 dakika cooldown
2. hata → 2 dakika cooldown
3. hata → 4 dakika cooldown
4. hata → 8 dakika cooldown
Max: 15 dakika
Basarili calisma → cooldown sifirla
YAPMA: Her hatada agent'i hemen tekrar calistir
YAP: Circuit breaker ile kontrol et
YAPMA: Hatalari sessizce yut
YAP: Logla, say, esik kontrolu yap
YAPMA: Tek hata tipine gore devre kes
YAP: Hata tipine gore agirlik ver
YAPMA: Sonsuz retry dongusu
YAP: Max retry + exponential backoff + fallback
YAPMA: Tum agent'lar icin ayni esik
YAP: Kritik agent'lar (security) icin dusuk esik
Integration with protocols.io API for managing scientific protocols. This skill should be used when working with protocols.io to search, create, update, or publish protocols; manage protocol steps and materials; handle discussions and comments; organize workspaces; upload and manage files; or integrate protocols.io functionality into workflows. Applicable for protocol discovery, collaborative protocol development, experiment tracking, lab protocol management, and scientific documentation.
Analyzes job descriptions and generates tailored resumes that highlight relevant experience, skills, and achievements to maximize interview chances
Generate Excalidraw diagrams from natural language descriptions. Use when asked to "create a diagram", "make a flowchart", "visualize a process", "draw a system architecture", "create a mind map", or "generate an Excalidraw file". Supports flowcharts, relationship diagrams, mind maps, and system architecture diagrams. Outputs .excalidraw JSON files that can be opened directly in Excalidraw.
Build and distribute Expo development clients locally or via TestFlight
Use when you have a written implementation plan to execute in a separate session with review checkpoints
Data structure for annotated matrices in single-cell analysis. Use when working with .h5ad files or integrating with the scverse ecosystem. This is the data format skill—for analysis workflows use scanpy; for probabilistic models use scvi-tools; for population-scale queries use cellxgene-census.
Benchling R&D platform integration. Access registry (DNA, proteins), inventory, ELN entries, workflows via API, build Benchling Apps, query Data Warehouse, for lab data management automation.
Comprehensive molecular biology toolkit. Use for sequence manipulation, file parsing (FASTA/GenBank/PDB), phylogenetics, and programmatic NCBI/PubMed access (Bio.Entrez). Best for batch processing, custom bioinformatics pipelines, BLAST automation. For quick lookups use gget; for multi-service integration use bioservices.
Take vibeeval/circuit-breaker from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.