Repository URL to install this package:
|
Version:
0.1.4 ▾
|
| agentio |
| agentio.egg-info |
| MANIFEST.in |
| PKG-INFO |
| README.md |
| setup.cfg |
| setup.py |
A trace visualization tool for the OpenAI Agents SDK, inspired by trackio.
pip install agentio
For additional features:
# Remote trace sending pip install agentio[remote] # Run as server pip install agentio[server] # Export to parquet pip install agentio[export]
import agentio from agents import Agent # Initialize local tracing agentio.init(project="my_assistant") # Your agent code - traces automatically captured agent = Agent(...) response = agent.run("Help me with...") # Export traces agentio.export("traces.json")
import agentio # Send traces to remote server agentio.init( project="production_bot", server_url="https://traces.mycompany.com", token="secret_token" ) # Traces automatically sent to server agent.run("...")
# Start trace collection server agentio serve --port 8080 --token my_secret_token
agentio serve - Run as trace collection serveragentio export - Export traces to fileagentio show - Display dashboard (UI not implemented yet)agentio.init()Initialize AgentIO tracing.
Parameters:
project (str): Project name for grouping tracesname (str, optional): Trace name (auto-generated if not provided)server_url (str, optional): Remote server URL for trace collectiondb_path (str, optional): Path to SQLite databasetoken (str, optional): Auth token for remote serverresume (str): Resume mode - "must", "allow", or "never"replace_default (bool): Replace OpenAI's default processoragentio.export()Export traces to file.
Parameters:
output_path (str): Output file pathformat (str): Export format - "json", "csv", or "parquet"project (str, optional): Project to exportagentio.serve()Run AgentIO as a standalone server.
Parameters:
host (str): Host to bind toport (int): Port to bind todb_path (str, optional): Database pathauth_token (str, optional): Authentication tokenMIT