references/openai-agents.md
AI agents: this is one page from PostHog's docs. Full index of Markdown docs for LLMs: https://posthog.com/llms.txt
OpenAI Agents SDK observability installation
Canonical URL: https://posthog.com/docs/ai-observability/installation/openai-agents
Cloud region: This page uses US Cloud hosts. For an EU Cloud project, use
https://eu.i.posthog.cominstead ofhttps://us.i.posthog.com(ingestion host), and usehttps://eu.posthog.cominstead ofhttps://app.posthog.com(app host for links).
On this page
- Install the PostHog SDK
- Install the OpenAI Agents SDK
- Initialize PostHog tracing
- Run your agents
- Multi-agent and tool usage
- Verify traces and generations
- Next steps
Let AI instrument your LLM calls for you
Skip the manual setup — run this in your project and the wizard installs the SDK and wires up AI Observability for you.
npx @posthog/wizard ai-observability
PostHog Wizard hedgehog
Let AI instrument your LLM calls for you
-
1
Install the PostHog SDK
Required
Setting up analytics starts with installing the PostHog Python SDK.
pip install posthog -
2
Install the OpenAI Agents SDK
Required
Install the OpenAI Agents SDK. PostHog instruments your agent runs by registering a tracing processor. The PostHog SDK does not proxy your calls.
pip install openai-agentsProxy note
These SDKs do not proxy your calls. They only fire off an async call to PostHog in the background to send the data. You can also use AI observability with other SDKs or our API, but you will need to capture the data in the right format. See the schema in the manual capture section for more details.
-
3
Initialize PostHog tracing
Required
Initialize PostHog with your project token and host from your project settings. Then call
instrument()to register PostHog tracing with the OpenAI Agents SDK. This automatically captures all agent traces, spans, and LLM generations.from posthog import Posthog from posthog.ai.openai_agents import instrument posthog = Posthog( "<ph_project_token>", host="https://us.i.posthog.com" ) instrument( client=posthog, distinct_id=lambda trace: (trace.metadata or {}).get("posthog_distinct_id"), privacy_mode=False, # optional groups={"company": "company_id_in_your_db"}, # optional )Note: If you want to capture LLM events anonymously, do not pass a distinct ID — here or per run. See our docs on anonymous vs identified events to learn more.
-
4
Run your agents
Required
Run your OpenAI agents as normal. PostHog automatically captures
$ai_generationevents for LLM calls and$ai_spanevents for agent execution, tool calls, and handoffs. Pass the user and conversation on the run'sRunConfig:group_idgroups the run's traces into a conversation — it becomes$ai_session_id.trace_metadata["posthog_distinct_id"]attributes the run's events to a user — thedistinct_idlambda from the previous step reads it off each trace. Any othertrace_metadatakeys land on the trace as$ai_trace_metadata.
The example below defines a tool and lets the agent call it.
from agents import Agent, Runner, RunConfig, function_tool @function_tool def get_weather(city: str) -> str: """Get the weather for a city.""" return f"The weather in {city} is sunny, 72F" agent = Agent( name="Assistant", instructions="You are a helpful assistant.", tools=[get_weather], ) result = Runner.run_sync( agent, "What's the weather in Paris?", run_config=RunConfig( group_id="conversation_abc", trace_metadata={"posthog_distinct_id": "user_123"}, ), ) print(result.final_output)You can expect captured
$ai_generationevents to have the following properties:| Property | Description | | --- | --- | |
$ai_model| The specific model, likegpt-5-miniorclaude-4-sonnet| |$ai_latency| The latency of the LLM call in seconds | |$ai_time_to_first_token| Time to first token in seconds (streaming only) | |$ai_tools| Tools and functions available to the LLM | |$ai_input| List of messages sent to the LLM | |$ai_input_tokens| The number of tokens in the input (often found in response.usage) | |$ai_output_choices| List of response choices from the LLM | |$ai_output_tokens| The number of tokens in the output (often found inresponse.usage) | |$ai_total_cost_usd| The total cost in USD (input + output) | | [...] | See full list of properties | -
5
Multi-agent and tool usage
Optional
PostHog captures the full trace hierarchy for complex agent workflows, including handoffs between multiple agents.
from agents import Agent, Runner, function_tool @function_tool def get_weather(city: str) -> str: """Get the weather for a city.""" return f"The weather in {city} is sunny, 72F" weather_agent = Agent( name="WeatherAgent", instructions="You help with weather queries.", tools=[get_weather] ) triage_agent = Agent( name="TriageAgent", instructions="Route weather questions to the weather agent.", handoffs=[weather_agent] ) result = Runner.run_sync(triage_agent, "What's the weather in San Francisco?")This captures:
- Agent spans for
TriageAgentandWeatherAgent - Handoff spans showing the routing between agents
- Tool spans for
get_weatherfunction calls - Generation spans for all LLM calls
As with the single-agent example above, PostHog captures every span in that list automatically. You write no extra code for the handoff itself.
- Agent spans for
-
Verify traces and generations
Recommended
Confirm LLM events are being sent to PostHog
Let's make sure LLM events are being captured and sent to PostHog. Under AI Observability, you should see rows of data appear in the Traces and Generations tabs.
LLM generations in PostHog
-
6
Next steps
Recommended
Now that you're capturing AI conversations, continue with the resources below to learn what else AI Observability enables within the PostHog platform.
| Resource | Description | | --- | --- | | Basics | Learn the basics of how LLM calls become events in PostHog. | | Generations | Read about the
$ai_generationevent and its properties. | | Traces | Explore the trace hierarchy and how to use it to debug LLM calls. | | Spans | Review spans and their role in representing individual operations. | | Anaylze LLM performance | Learn how to create dashboards to analyze LLM performance. |