vela-sdk 0.1.1

Observability and runtime governance for AI agent pipelines

Vela

See inside your AI pipelines. Know what's slow, what's failing, and what's costing you money.

Install

# Create and activate a virtual environment
python3 -m venv venv
source venv/bin/activate

# Install vela-sdk
pip3 install vela-sdk

On Python 3.11+, virtual environments are required due to system protection. On Windows, use venv\Scripts\activate instead.

Get your API key

https://dashboard-one-tau-97.vercel.app/keys

Instrument your pipeline

import vela

vela.init(api_key='your-key-here')

@vela.trace
def researcher(topic):
    # your existing code — unchanged
    ...

@vela.trace
def writer(research):
    # your existing code — unchanged
    ...

vela.new_session()
result = researcher('your topic')
output = writer(result)
vela.end_session()

See your traces

https://dashboard-one-tau-97.vercel.app?key=your-key-here

That's it. No other changes to your code.

Optional: label your sessions

vela.new_session(label='production run')

Optional: track model and cost

@vela.trace(name='researcher', model='gpt-4o')
def researcher(topic):
    ...