Metadata-Version: 2.4
Name: vantio-agent-sdk
Version: 2.0.0
Summary: Vantio AI Zero-Trust Execution Interceptor
Requires-Python: >=3.8
Description-Content-Type: text/markdown
Requires-Dist: requests>=2.25.0

# Vantio Enterprise AI Agent SDK (Python)

Strictly opt-in observability tooling for enterprise AI workloads.

## Overview

This SDK provides a lightweight session-boundary layer for AI API traffic
tracing. It is a developer-configured utility — it does **nothing** unless
you explicitly instantiate and enter a `VantioSession` context.

## Configuration

Two environment variables must be present before the SDK is meaningful in your
deployment:

| Variable | Purpose |
|---|---|
| `VANTIO_API_KEY` | Your Vantio enterprise credential, used to authenticate with the Vantio control plane. |
| `VANTIO_PROXY_ENDPOINT` | The endpoint your AI traffic is routed through. No proxy is active without this value. |

## Usage

```python
from vantio.session import VantioSession

with VantioSession(agent_name="my-agent") as session:
    # Within this block, VANTIO_TRACE_ID is set in the process environment.
    # Your AI API calls made here are associated with session.trace_id.
    print("Trace ID:", session.trace_id)
    # ... make AI API calls ...
```

When engaged, the SDK prints a read-only trace link to the terminal for
real-time debugging. The `VANTIO_TRACE_ID` environment variable is set only
for the duration of the `with` block and is removed on exit.

## Enterprise opt-in model

- Installation alone produces no side effects.
- No network connections are made by the SDK itself.
- Session tracing is activated only by explicitly entering a `VantioSession`
  context.
- Developers retain full control of which code paths are instrumented.
