AI Operations Utilities

A collection of lightweight, reusable Python helpers built and maintained by the Snyk AI Operations Team.

---

✨ Features

Area	Module	What it does
Retrieval	aiops_utils.retrievers.SnykMultiSourceRetriever	LangChain-compatible retriever that queries a remote multi-source search service, merges the results, then returns them as langchain_core.Document objects.

---

📦 Installation

pip install git+https://github.com/Snyk-CIS/aiops-utils.git

Python ≥ 3.10 is required.

---

🚀 Quick-start

1. Retrieve Documents with LangChain

from aiops_utils.retrievers import SnykMultiSourceRetriever

retriever = SnykMultiSourceRetriever(
    jwt_token="your_jwt_token",
    app_name="your_app_name",
    service_names=["SOURCE_A", "SOURCE_B"],
    service_max_documents={"SOURCE_A": 5, "SOURCE_B": 3},
    service_confidence_thresholds={"SOURCE_A": 0.9, "SOURCE_B": 1.0},
    service_filters={"SOURCE_B": {"@eq": {"author": "example_user"}}},
    user_email="[email protected]",
    grading=True,
    decomposition=True
)

results = retriever.invoke("How do I rotate my credentials?")
for doc in results:
    print(doc.page_content)

Key constructor flags:

Parameter	Required	Purpose
jwt_token	✅	Auth bearer token passed in the Authorization header.
app_name	✅	Name of the search service (used for DNS discovery).
service_names	✅	Either 'all' or a list of specific back-end sources.
service_max_documents	❌	Per-source override for maximum number of documents to return.
service_confidence_thresholds	❌	Per-source override for minimum confidence scores.
service_filters	❌	Dictionary mapping service names to filter objects.
grading	❌	When True, the backend returns per-token confidence scores. Defaults to None.
decomposition	❌	When True, enables query decomposition to break complex queries into subproblems. Auto-enables grading. Defaults to None.
user_email	❌	Email address of the user making the request. Used for tracking. Defaults to None.

---

🛠️ Development

For maintainers and contributors working on the aiops-utils package.

Building the Package

Use the provided script to rebuild the package:

./scripts/rebuild_package.sh

This script will:

• ✅ Check your environment
• 🔢 Optionally increment the version
• 🧹 Clean old build artifacts
• 🔨 Build wheel and source distributions
• 📋 Show next steps for deployment

Manual Build Steps

If you prefer manual control:

# 1. Update version in pyproject.toml (if needed)
# 2. Clean and build
rm -rf dist/ build/ src/aiops_utils.egg-info/
pip install --upgrade build wheel setuptools
python -m build

# 3. Commit and push
git add . && git commit -m "Release version X.X.X" && git push

Deployment to Production

1. Push to Git (apps install from Git, not PyPI):

   git push origin main

2. Production apps will automatically get the latest version on their next deployment

3. Force update existing deployments (if needed):

   pip install --upgrade --force-reinstall git+https://github.com/Snyk-CIS/aiops-utils.git