Metadata-Version: 2.4
Name: iris-langfuse
Version: 0.1.0
Summary: IRIS compliance intelligence from Langfuse traces — keep your observability stack, add governance
Author-email: IRIS Platform <sdk@iris.ai>
License: Apache-2.0
Project-URL: Homepage, https://iris-security.github.io/iris-sdk/
Project-URL: Repository, https://github.com/IRIS-Security/iris-sdk
Project-URL: Documentation, https://iris-security.github.io/iris-sdk/getting-started.html
Project-URL: Issues, https://github.com/IRIS-Security/iris-sdk/issues
Requires-Python: >=3.10
Description-Content-Type: text/markdown
Requires-Dist: iris-security-sdk>=0.2.16
Requires-Dist: httpx>=0.27
Provides-Extra: live
Requires-Dist: langfuse<4,>=2; extra == "live"
Provides-Extra: dev
Requires-Dist: pytest>=8.0; extra == "dev"
Requires-Dist: ruff>=0.4; extra == "dev"

# iris-langfuse

Read your existing [Langfuse](https://langfuse.com) project and derive an IRIS **WorkloadProfile** for compliance intelligence — without reimplementing tracing.

**Keep your Langfuse setup. IRIS reads what you're running and tells you which regulations apply — with tamper-evident proof.**

## Install

```bash
pip install "iris-langfuse[live]"
```

## Quickstart

```python
from iris_langfuse import profile_from_langfuse

profile = profile_from_langfuse(lookback_days=30)
print(profile["models"], profile["providers"])
```

Set `LANGFUSE_PUBLIC_KEY`, `LANGFUSE_SECRET_KEY`, and optionally `LANGFUSE_HOST`.

## Privacy

IRIS only reads trace **names, tags, and metadata** — never prompt or output content.

## IRIS Cloud bridge

```bash
export IRIS_API_KEY=your-token
python examples/governed_langfuse.py --push
```

Optional POST to `/intelligence/profile/scan` when `IRIS_API_KEY` is set.

## See also

- [Integrations docs](https://iris-security.github.io/iris-sdk/integrations.html)
- [Compliance Intelligence](https://iris-security.github.io/iris-sdk/compliance-intelligence.html)
