TraceAI BeeAI Framework Integration

Comprehensive observability for IBM BeeAI Framework with TraceAI.

BeeAI Framework is an open-source toolkit from IBM Research for building production-grade multi-agent systems. This integration provides seamless tracing by leveraging BeeAI's built-in OpenInference instrumentation.

Features

• Zero-config integration: Works with BeeAI's native OpenInference support
• Automatic tracing: Agent runs, tool calls, LLM interactions, and workflows
• Token tracking: Input/output tokens and usage metrics
• Session correlation: Link traces across conversations
• Custom middleware: Extended event capture for detailed observability
• IBM Granite support: Optimized for IBM Granite and Llama models

Installation

pip install traceai-beeai

For full functionality with BeeAI:

pip install beeai-framework openinference-instrumentation-beeai

Quick Start

IMPORTANT: Call configure_beeai_tracing() BEFORE importing BeeAI modules!

Option 1: Using TraceAI with fi_instrumentation

# Setup TraceAI FIRST
from fi_instrumentation import register
from fi_instrumentation.fi_types import ProjectType
from traceai_beeai import configure_beeai_tracing

trace_provider = register(
    project_type=ProjectType.OBSERVE,
    project_name="my-beeai-agent",
)

configure_beeai_tracing(tracer_provider=trace_provider)

# NOW import BeeAI modules
from beeai_framework.agents import Agent
from beeai_framework.backend.chat import ChatModel

model = ChatModel.from_name("ollama:granite3.1-dense:8b")
agent = Agent(
    llm=model,
    role="Assistant",
    instructions="You are a helpful assistant.",
)

response = agent.run("Hello!")

Option 2: Direct OTLP Configuration

from traceai_beeai import configure_beeai_tracing

# Configure with OTLP endpoint directly
configure_beeai_tracing(
    otlp_endpoint="https://api.traceai.com/v1/traces",
    otlp_headers={"Authorization": "Bearer YOUR_API_KEY"},
    project_name="my-beeai-agent",
)

# Now import and use BeeAI
from beeai_framework.agents import Agent
# ...

Using Tools

BeeAI tools are automatically traced:

from beeai_framework.agents import Agent
from beeai_framework.backend.chat import ChatModel
from beeai_framework.tools import WikipediaTool, OpenMeteoTool, ThinkTool

model = ChatModel.from_name("ollama:granite3.1-dense:8b")

agent = Agent(
    llm=model,
    role="Research Assistant",
    instructions="Use tools to find information.",
    tools=[
        ThinkTool(),       # Internal reasoning
        WikipediaTool(),   # Knowledge retrieval
        OpenMeteoTool(),   # Weather forecasts
    ],
)

response = agent.run("What's the weather in Paris?")

Custom Middleware

For extended tracing with session tracking:

from traceai_beeai import create_tracing_middleware

middleware = create_tracing_middleware(
    tracer_provider=trace_provider,
    capture_input=True,
    capture_output=True,
    session_id="user-session-123",
    user_id="user@example.com",
)

agent = Agent(
    llm=model,
    role="Assistant",
    instructions="You are helpful.",
    middlewares=[middleware],
)

Using Requirements

BeeAI agents with requirements are fully traced:

from beeai_framework.agents import Agent
from beeai_framework.requirements import ConditionalRequirement
from beeai_framework.tools import ThinkTool

agent = Agent(
    llm=model,
    role="Safe Assistant",
    instructions="Always think before responding.",
    tools=[ThinkTool()],
    requirements=[
        ConditionalRequirement(
            step=0,
            tool=ThinkTool,  # Force ThinkTool on first step
        ),
    ],
)

Semantic Attributes

The integration captures these OpenTelemetry GenAI semantic attributes:

Agent Attributes

Attribute	Description
agent.name	Agent name
agent.type	Agent class name
agent.role	Agent role description
agent.instructions	Agent instructions (truncated)
beeai.tool_count	Number of tools available
beeai.requirements	Configured requirements
beeai.memory.type	Memory strategy type

Model Attributes

Attribute	Description
gen_ai.system	Model provider (ibm, openai, etc.)
gen_ai.request.model	Model name/ID
gen_ai.request.temperature	Temperature setting
gen_ai.request.max_tokens	Max tokens setting
gen_ai.usage.input_tokens	Input token count
gen_ai.usage.output_tokens	Output token count

Tool Attributes

Attribute	Description
gen_ai.tool.name	Tool name
gen_ai.tool.description	Tool description
gen_ai.tool.parameters	Tool input parameters
gen_ai.tool.result	Tool execution result

Workflow Attributes

Attribute	Description
beeai.workflow.name	Workflow name
beeai.workflow.step	Current step name
beeai.session.id	Session identifier
beeai.user.id	User identifier

Model Provider Support

The integration automatically detects model providers:

Provider	Model Patterns
IBM	granite-*, ibm/*
OpenAI	gpt-*, o1-*
Anthropic	claude-*
Meta	llama-*
Google	gemini-*
Mistral	mistral-*, mixtral-*
Ollama	ollama/*
Groq	groq/*
Together	together/*

Environment Variables

BeeAI respects standard OpenTelemetry environment variables:

# OTLP endpoint
export OTEL_EXPORTER_OTLP_TRACES_ENDPOINT="https://api.traceai.com/v1/traces"

# OTLP headers (authentication)
export OTEL_EXPORTER_OTLP_HEADERS="Authorization=Bearer YOUR_API_KEY"

Examples

See the examples/ directory for complete examples:

• basic_agent.py - Simple agent with tracing
• agent_with_tools.py - Tools and function calling
• requirements_agent.py - Behavioral requirements
• middleware_tracing.py - Custom middleware with session tracking

How It Works

BeeAI has native observability support via OpenInference instrumentation. This integration:

1. Configures OpenTelemetry: Sets up tracer provider and OTLP export
2. Initializes BeeAIInstrumentor: Wraps BeeAI internals for automatic span creation
3. Provides middleware: Optional TraceAIMiddleware for extended capture
4. Adds helper functions: Attribute extraction and trace context management

Compatibility

• BeeAI Framework: >= 0.1.0
• Python: >= 3.11
• OpenTelemetry: >= 1.0.0
• OpenInference: >= 0.1.0

Resources

• BeeAI Framework Documentation
• BeeAI GitHub
• IBM Research - BeeAI
• OpenTelemetry GenAI Conventions
• TraceAI Documentation

License

Apache-2.0