Agno Telemetry

When telemetry is enabled, Agno sends per-event usage metadata to https://os-api.agno.com by default. Agents, teams, and workflows send an event for each run. Evals send an event for each evaluation, and AgentOS sends an event when it launches.

Telemetry payloads do not include prompts or responses. They include raw component, session, and run IDs. Parser and output model configuration is serialized with each model’s to_dict() method, which can include user-supplied request fields such as extra_headers, extra_query, or extra_body. Do not place credentials or other sensitive data in those fields or IDs when telemetry is enabled.

Data Collected

Event	Metadata
Agent run	Session, run, and agent IDs; SDK version; database class; model identifiers; feature flags
Team run	Session, run, and team IDs; SDK version; model identifiers; member count; feature flags
Workflow run	Session, run, and workflow IDs; SDK version; database class; input-schema flag
Eval run	Eval run ID and type; SDK version; model or component identifiers; evaluation configuration counts and flags
AgentOS launch	AgentOS ID; SDK version; registered agent, team, and workflow IDs; interface types

Example Telemetry Payload

An agent run telemetry payload has this shape:

{
    "session_id": "123",
    "run_id": "123",
    "sdk_version": "2.7.4",
    "type": "agent",
    "data": {
        "agent_id": "123",
        "db_type": "PostgresDb",
        "model_provider": "OpenAI",
        "model_name": "OpenAIResponses",
        "model_id": "gpt-5.2",
        "parser_model": {
            "name": "OpenAIResponses",
            "id": "gpt-5.2",
            "provider": "OpenAI"
        },
        "output_model": {
            "name": "OpenAIResponses",
            "id": "gpt-5.2",
            "provider": "OpenAI"
        },
        "has_tools": true,
        "has_memory": false,
        "has_learnings": false,
        "has_culture": false,
        "has_reasoning": true,
        "has_knowledge": true,
        "has_input_schema": false,
        "has_output_schema": false,
        "has_team": true
    }
}

How to Disable Telemetry

You can disable telemetry in two ways:

Environment Variable

Set AGNO_TELEMETRY to false to disable telemetry for agents, teams, and workflows:

export AGNO_TELEMETRY=false

AgentOS and evals do not read AGNO_TELEMETRY. Configure those instances directly with telemetry=False.

Per-Instance Configuration

Disable telemetry for specific agents, teams, workflows, or AgentOS instances:

from agno.agent import Agent
from agno.models.openai import OpenAIResponses

# Disable telemetry for a specific agent
agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),
    telemetry=False
)

This works for:

Agents: Agent(telemetry=False)
Teams: Team(telemetry=False)
Workflows: Workflow(telemetry=False)
Evals: AccuracyEval(telemetry=False) (also PerformanceEval, ReliabilityEval, and AgentAsJudgeEval)
AgentOS: AgentOS(telemetry=False)

Get Started

Basics

Advanced

Production

Providers

Other

Additional Resources

Agno Telemetry

Data Collected

Example Telemetry Payload

How to Disable Telemetry

Environment Variable

Per-Instance Configuration

Developer Resources

​Data Collected

​Example Telemetry Payload

​How to Disable Telemetry

​Environment Variable

​Per-Instance Configuration

​Developer Resources

Data Collected

Example Telemetry Payload

How to Disable Telemetry

Environment Variable

Per-Instance Configuration

Developer Resources