Trace and evaluate Strands agents with Future AGI

Community Contribution

This is a community-maintained package that is not owned or supported by the Strands team. Validate and review the package before using it in your project. Have your own integration? We'd love to add it here too!

Future AGI is an open-source e2e agent engineering and optimization platform that helps you ship self-improving AI agents. Strands traces export over OpenTelemetry, so they can be sent to the Future AGI tracer endpoint without any code changes to your agent.

This page walks through two setup paths:

1. The recommended traceai-strands package, which wires up the project resource attributes Future AGI uses to bucket spans into a named project.
2. A manual setup using only Strands’ built-in OpenTelemetry support and OTLP environment variables.

Prerequisites

Sign up at app.futureagi.com and copy your FI_API_KEY and FI_SECRET_KEY from the dashboard.

export FI_API_KEY="your-fi-api-key"
export FI_SECRET_KEY="your-fi-secret-key"

Option 1: Recommended setup with traceai-strands

Install Strands with OpenTelemetry support and the Future AGI integration package:

pip install 'strands-agents[otel]' traceai-strands

Register the tracer once at startup, then attach it to Strands. The register() call sets all the resource attributes Future AGI needs to surface the project in the dashboard.

from fi_instrumentation import register
from fi_instrumentation.fi_types import ProjectType
from traceai_strands import configure_strands_tracing

trace_provider = register(
    project_type=ProjectType.OBSERVE,
    project_name="my-strands-agent",
)
configure_strands_tracing(tracer_provider=trace_provider)

from strands import Agent

agent = Agent(
    model="us.anthropic.claude-sonnet-4-20250514-v1:0",
    system_prompt="You are a helpful assistant.",
)

response = agent("Hello!")

Open your project in the Future AGI dashboard to inspect the run tree, prompts, completions, model parameters, token usage, and tool execution details.

Option 2: Manual setup with StrandsTelemetry

If you would rather not add another package, you can configure Strands’ built-in OTLP exporter directly. Future AGI’s tracer endpoint accepts standard OTLP/HTTP at https://api.futureagi.com/tracer/v1/traces.

export OTEL_EXPORTER_OTLP_ENDPOINT="https://api.futureagi.com/tracer/v1/traces"
export OTEL_EXPORTER_OTLP_HEADERS="x-api-key=${FI_API_KEY},x-secret-key=${FI_SECRET_KEY}"
export OTEL_RESOURCE_ATTRIBUTES="project_name=my-strands-agent,project_type=observe,project_version_name=DEFAULT"

from strands.telemetry import StrandsTelemetry

StrandsTelemetry().setup_otlp_exporter()

from strands import Agent

agent = Agent(
    model="us.anthropic.claude-sonnet-4-20250514-v1:0",
    system_prompt="You are a helpful assistant.",
)

response = agent("Hello!")

The project_name, project_type, and project_version_name resource attributes are required for Future AGI to bucket spans into a named project — without them, traces are accepted but not surfaced in the dashboard. Option 1 sets these automatically.

What gets traced

Once configured, every agent invocation, model call, tool execution, and event-loop cycle is captured as an OpenTelemetry span and forwarded to Future AGI. See the Strands traces page for the full attribute schema.

Resources

• traceai-strands on PyPI
• Future AGI Strands integration guide