Overview / Dashboard
workspace: production
OVERVIEW
// real-time platform status · last updated just now
Active Agents
6
↑ 2 from yesterday
Requests Today
2,847
↑ 12% vs last week
Avg Latency
1.4s
↓ 0.3s improved
Token Cost Today
$3.2
↑ $0.8 budget 80%
Live Request Flow
● 14 req/min
👤User / App
Model Router
🔧MCP Server
⚙️Pool Manager
🤖Agent Pool
📊Response
🔀 Model Router Usage
🟠 Anthropic Claude58%
🔵 Azure OpenAI26%
🟢 AWS Bedrock16%
📋 Recent Activity
12:43:01 OK sap-agent Fetched 48 purchase orders from SAP
12:42:38 INFO research-agent Web search completed — 12 results
12:41:55 WARN router Claude rate limit, fallback to Azure
12:40:12 OK data-agent SQL query executed — 2,341 rows
AGENT POOL
// manage, configure, and deploy AI agents
🔍
Research Agent
research-agent-v2
● active
search_web read_file summarize
842
runs
1.2s
avg
98%
success
💻
Coding Agent
coding-agent-v3
● busy
code_exec git_tools file_write
421
runs
3.8s
avg
94%
success
🏭
SAP Agent
sap-erp-agent-v1
● active
sap_odata query_db report_gen
1,204
runs
2.1s
avg
99%
success
📊
Data Agent
data-analyst-v2
● active
sql_query analytics visualize
589
runs
4.2s
avg
96%
success
📬
Notif Agent
notification-agent-v1
● active
send_email webhook slack_msg
3,102
runs
0.4s
avg
99.9%
success
Custom Agent
new-agent
○ draft
Define a new agent using YAML config or the visual editor. Add tools, model, system prompt and deploy.
MODEL ROUTER
// LiteLLM-powered · auto failover · cost optimization
🔀 Configured Models
🟠
Anthropic Claude Sonnet
claude-sonnet-4-5 · priority: 1 · reasoning, coding
load58%
● primary
🔵
Azure OpenAI GPT-4o
gpt-4o · priority: 2 · general, fallback
load26%
● fallback
🟢
AWS Bedrock Claude Haiku
claude-haiku · priority: 3 · fast, cheap
load16%
● cost-save
⚙️ Routing Rules
Complex Reasoning
IF task_type == "reasoning" → USE claude-sonnet
Cost Mode
IF cost_mode == "cheap" → USE claude-haiku (bedrock)
Auto Failover
IF primary_fails → FALLBACK to next_priority
Rate Limit Guard
IF rate_limit_hit → ROTATE to standby model
MCP TOOLS
// Model Context Protocol server · tool registry
🏭
SAP OData API
tool::sap_odata
Query SAP ERP via OData V2/V4. Supports purchase orders, materials, sales, FICO reports.
🗄️
Database Query
tool::query_db
Execute read/write SQL queries to PostgreSQL and MySQL. Supports parameterized queries.
🌐
Web Search
tool::search_web
Real-time web search via Bing/Google API. Returns top results with snippets and metadata.
📁
File System
tool::read_file
Read, write, list files from server filesystem or S3/GCS bucket. Supports PDF parsing.
📨
Notification
tool::send_notif
Send alerts via Email, Slack, Telegram, or custom Webhook. Supports rich formatting.
Add Custom Tool
tool::custom
Register any REST/OData/gRPC endpoint as an MCP tool. Define schema via YAML or JSON.
RUN LOGS
// real-time execution log · all agents & tools
TIME LEVEL AGENT MESSAGE LATENCY
CHATBOT
// test agents interactively · live execution
Select Agent
Model Override
Active MCP Tools
search_web
read_file
sap_odata
query_db
Last Run Stats
Latency1.24s
Tokens842
Model usedclaude-sonnet
Tools called2
🤖
Halo! Saya Research Agent. Saya bisa mencari informasi, membaca file, dan merangkum konten untuk kamu. Mau mulai dari mana?
research-agent-v2 · claude-sonnet · just now