Metadata-Version: 2.4
Name: sectum-ai-probes
Version: 0.1.8
Summary: Sectum AI - the multi-tenant leakage attack catalog (probe classes).
Project-URL: Homepage, https://sectum.ai
Project-URL: Repository, https://github.com/sectum-ai/sectum-ai
Author: Sectum AI
License-Expression: Apache-2.0
Keywords: ai-security,multi-tenant,red-team,verification
Classifier: Development Status :: 2 - Pre-Alpha
Classifier: Intended Audience :: Developers
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Security
Requires-Python: >=3.12
Requires-Dist: pyyaml>=6
Requires-Dist: sectum-ai-adapters
Requires-Dist: sectum-ai-spec
Description-Content-Type: text/markdown

# sectum-ai-probes

The multi-tenant leakage **attack catalog** and leak-detection pipeline for
[Sectum AI](https://github.com/sectum-ai/sectum-ai).

This distribution implements the Class 1–11 probes — direct tenant-boundary
fetch, the flagship organic entity-bleed RAG (the 95.4% Retrieval-Pivot
finding), adversarial RAG poisoning, semantic-cache contamination, KV-cache
timing, embedding inversion, agent / MCP tool-call hijacking, persistent-memory
contamination, LoRA cross-tenant influence, IKEA-style benign extraction, and
the GDPR Article 17 erasure-verification wedge — behind a single pluggable
`Probe` interface, plus the exact → semantic → calibrated-judge detection
pipeline that turns observations into manifest-grounded `Finding`s.

```sh
pip install sectum-ai-probes
```

Most users install the umbrella package [`sectum-ai`](https://pypi.org/project/sectum-ai/)
instead, which pulls this in automatically.

- Attack catalog (one page per class): <https://docs.sectum.ai>
- Source: <https://github.com/sectum-ai/sectum-ai>

Apache-2.0.
