Traces (12)
Call Tree — trace_abc123
Node Details
Select a trace node to view details
8
Active Alerts
3
Critical
4
Warning
1
Info
Cost by Agent (30d)
Token Usage by Model
GPT-4o
65%
GPT-4o-mini
30%
Claude 3.5
5%
Total Tokens (30d)
847.3M
Avg per Task
12,847
Cost Trend (30 Days)
$8
$487.50 this month
$22
Agent Version Comparison
Top Failure Modes
1
Tool timeout exceeded
127
2
Context window overflow
93
3
Rate limit hit (API)
67
4
Invalid tool response
42
5
Auth token expired
23
Token Statistics
Total Input
512.4M
Total Output
334.9M
Avg Input/Task
7,812
Avg Output/Task
5,035
Cost/1M Input
$2.50
Cost/1M Output
$10.00
Optimization Recommendations
Switch to GPT-4o-mini for code_review
Your code review agent is using GPT-4o but outputs are simple diffs. GPT-4o-mini would reduce costs by 73% with negligible quality impact.
Add response caching layer
18% of your research_assistant queries are identical (repeated FAQ lookups). A semantic cache could save $31/month.
Investigate data_pipeline timeouts
data_pipeline has 23 tool timeouts in the last 24h. The database queries are likely exceeding the 30s timeout threshold.
API Key
API Key
sk_agentscope_••••••••••••••••3f9a
SDK Connection
pip install agentscope-sdk
from agentscope import AgentScope
client = AgentScope(
api_key="sk_agentscope_•••••••••",
project="production"
)
# Auto-instrument your LangChain agent
agent = client.wrap_agent(your_existing_agent)
Framework Adapters
LangChain
Connected — v0.2.12
CrewAI
Connected — v0.30.0
OpenAI Agents SDK
Connected — v1.0.4
Integrations
Slack
Not connected
PagerDuty
Connected — alerts@agentscope
Webhook URL
Data Retention
Trace Retention
Metric Retention
Log Storage
Account
Plan
Pro
Billing Cycle
Monthly
Monthly Limit
50,000 traces
Usage This Month
2,159 / 50,000 (4%)
Security
Two-Factor Auth
Audit Logs
IP Allowlist