moduna 0.1.9

Moduna Python SDK for tracing AI agent frameworks with Traceloop.

Moduna Python SDK

Moduna is a small Python SDK that brands and simplifies Traceloop setup for AI agent frameworks. It gives applications a stable moduna import path, framework-aware instrumentation, and Moduna API-key resolution from configuration, exported environment variables, or .env.

Installation

Install the SDK with all supported framework integrations:

uv add moduna
pip install moduna

The framework extras are also available for teams that prefer explicit dependency declarations:

uv add "moduna[langchain]"
pip install "moduna[langchain]"

uv add "moduna[crewai]"
pip install "moduna[crewai]"

Quick Start

from moduna import Instruments, Moduna

moduna = Moduna()
moduna.init(
    {
        "app_name": "customer support",
        "framework": Instruments.LANGCHAIN,
        "api_key": "mod_...",
    }
)

Set a conversation ID using the Moduna-branded tracing namespace:

from moduna.sdk.tracing import set_conversation_id

set_conversation_id("customer-123")

Configuration

Moduna resolves the API key in this order:

1. api_key passed to Moduna().init(...)
2. exported MODUNA_API_KEY
3. MODUNA_API_KEY=... in a local .env file

The production Moduna endpoint is used by default. For local development, create an ignored config.toml in the project root to switch to a local collector:

[sdk]
base_url = "http://localhost:4318"

You can also override the endpoint per application:

from moduna import Instruments, Moduna

Moduna().init(
    {
        "app_name": "support-agent",
        "framework": Instruments.LANGCHAIN,
        "base_url": "http://localhost:4318",
    }
)

Frameworks

LangChain

from langchain_core.prompts import ChatPromptTemplate
from moduna import Instruments, Moduna
from moduna.sdk.tracing import set_conversation_id

Moduna().init({"app_name": "langchain-app", "framework": Instruments.LANGCHAIN})
set_conversation_id("conversation-1")

prompt = ChatPromptTemplate.from_messages([("human", "{question}")])

CrewAI

from moduna import Instruments, Moduna
from moduna.sdk.tracing import set_conversation_id

Moduna().init({"app_name": "crewai-app", "framework": Instruments.CREWAI})
set_conversation_id("crew-run-1")

Moduna installs the Traceloop/OpenTelemetry CrewAI instrumentation. Install the CrewAI runtime separately in your application when you build CrewAI agents.

Development

Activate the local virtual environment before running checks:

source .venv/bin/activate
uv sync
uv run ruff format .
uv run ruff check .
uv run pytest
uv build

Use uv run ruff format --check . in CI when you want formatting verification without rewriting files.

Publishing

GitHub Actions validates pull requests and every push to main. A successful main build publishes only when the version in pyproject.toml does not already exist on PyPI.

Configure the PyPI token once:

1. Open the GitHub repository.
2. Go to Settings → Secrets and variables → Actions.
3. Add a repository secret named PYPI_TOKEN.
4. Paste the PyPI token value from your local .env.

Never commit .env or the token.

Prepare a release with uv:

source .venv/bin/activate
uv version --bump patch
git add pyproject.toml uv.lock
git commit -m "Release Moduna $(uv version --short)"
git push origin main

Changes that do not bump the version still run the full CI pipeline and skip the PyPI upload successfully.