Build Your Agent
Agent Templates
Pre-built agents ready to deploy. Click one to load it into the builder.
Agent Chat
Branching Chat
Explore "what if" scenarios by forking conversations. Click the fork icon on any message to create a branch.
Live Events
Cost Over Time
Most Used Tools
No tool usage recorded yet. Run an agent with tools enabled.
Model Comparison
| Model | Calls | Total Cost | Avg Cost | Avg Latency | Avg Tokens | Total Tokens |
|---|---|---|---|---|---|---|
| No model data yet. | ||||||
Agent Leaderboard
| # | Agent | Quality | Queries | Tool Calls | Total Cost | Cost/Query | Events |
|---|---|---|---|---|---|---|---|
| No agent data yet. | |||||||
Loading marketplace...
Publish Your Agent
Share your agent configuration with the community.
Embeddable Chat Widget
Generate an embeddable widget so other companies can add AgentOS agents to their websites with a single script tag.
Python SDK
Use the Python SDK to integrate AgentOS into any backend:
from agentos.embed import AgentOSClient
client = AgentOSClient(
base_url="http://localhost:8000",
api_key="your-api-key",
)
# Single response
response = client.run("How can I help?")
print(response)
# Streaming
for token in client.stream("Tell me a story"):
print(token, end="", flush=True)
# Browse marketplace
agents = client.list_agents()
for a in agents:
print(a["name"], a["downloads"])
Agent-to-Agent Mesh
Discover, authenticate, negotiate, and transact with agents across organisations using the mesh protocol.
Registry
Agents registered in the mesh network.
Register Agent
Add a new agent to the mesh network.
Send Message
Send a mesh protocol message to another agent.
Transactions
Ledger of all mesh transactions.
Message Log
Agent Simulation World
Stress-test your agent with realistic simulated customers. Generate 50-100 concurrent interactions from different personas, then review quality scores and failure analysis.
Configure Simulation
Persona Breakdown
Score Distribution
Failure Analysis
Worst Interactions
Agent Learning System
Collect user feedback, analyze failure patterns, auto-optimise prompts, and build few-shot examples — all without fine-tuning.
Submit Feedback
Topic Analysis
Which topics does the agent handle well (or poorly)?
Prompt Patches
Auto-generated prompt improvements for weak areas.
Few-Shot Examples
Best interactions auto-selected as in-context examples.
Learning Progress
Recent Feedback
Root Cause Analysis
Deep tracing, 5-point diagnostics, smart causal alerts, and step-by-step replay of agent interactions.
Smart Alerts
Causal alerts that explain WHY something is wrong.
Recent Traces
Interaction Replay
Click a failed trace above to replay it step-by-step.
5-Point Diagnosis
Agent Scheduler
Schedule agents to run automatically at intervals or cron times.
Active Jobs
No scheduled jobs. Create one above.
Event Bus
Fire events and see which agents react. Supports webhooks, timers, agent-to-agent chains, and custom events.
Send JSON body to fire events from external services
Registered Listeners
No listeners registered. Use the Python API to register agents.
Event History
No events emitted yet.
Account
Register or log in to get an API key. This key is used to track your usage.
API key is stored locally in your browser (localStorage) and sent as X-API-Key for API calls.
Your Usage
Log in to see your usage.
A/B Testing
Compare two agent configurations on the same set of queries using an LLM-as-judge.
Agent A
Agent B
Results
No A/B test run yet.
Multi-modal Analysis
Upload an image or document and ask questions about it. Supports images (PNG, JPG, GIF, WebP) and documents (TXT, MD, PDF, CSV, JSON).
Click to upload or drag & drop
Images: PNG, JPG, GIF, WebP · Documents: TXT, MD, PDF, CSV, JSON
Or Analyze by URL
Paste a public image URL to analyze without uploading.