Metadata-Version: 2.2
Name: llm-security-gates
Version: 0.1.1
Summary: Practical, FOSS-backed security gates for LLM systems: model-file supply-chain gate, runtime I/O firewall proxy, and model resilience scoring.
Author: Gexiro Global Enterprises Ltd
License: Apache License
        Version 2.0, January 2004
        http://www.apache.org/licenses/
        
        Copyright 2026 Gexiro Global Enterprises Ltd.
        
        TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
        
        1. Definitions.
        
        "License" shall mean the terms and conditions for use, reproduction, and
        distribution as defined by Sections 1 through 9 of this document.
        
        "Licensor" shall mean the copyright owner or entity authorized by the copyright
        owner that is granting the License.
        
        "Legal Entity" shall mean the union of the acting entity and all other entities
        that control, are controlled by, or are under common control with that entity.
        For the purposes of this definition, "control" means (i) the power, direct or
        indirect, to cause the direction or management of such entity, whether by
        contract or otherwise, or (ii) ownership of fifty percent (50%) or more of the
        outstanding shares, or (iii) beneficial ownership of such entity.
        
        "You" (or "Your") shall mean an individual or Legal Entity exercising
        permissions granted by this License.
        
        "Source" form shall mean the preferred form for making modifications, including
        but not limited to software source code, documentation source, and configuration
        files.
        
        "Object" form shall mean any form resulting from mechanical transformation or
        translation of a Source form, including but not limited to compiled object code,
        generated documentation, and conversions to other media types.
        
        "Work" shall mean the work of authorship, whether in Source or Object form,
        made available under the License, as indicated by a copyright notice that is
        included in or attached to the work.
        
        "Derivative Works" shall mean any work, whether in Source or Object form, that
        is based on (or derived from) the Work and for which the editorial revisions,
        annotations, elaborations, or other modifications represent, as a whole, an
        original work of authorship. For the purposes of this License, Derivative Works
        shall not include works that remain separable from, or merely link (or bind by
        name) to the interfaces of, the Work and Derivative Works thereof.
        
        "Contribution" shall mean any work of authorship, including the original version
        of the Work and any modifications or additions to that Work or Derivative Works
        thereof, that is intentionally submitted to Licensor for inclusion in the Work
        by the copyright owner or by an individual or Legal Entity authorized to submit
        on behalf of the copyright owner. For the purposes of this definition,
        "submitted" means any form of electronic, verbal, or written communication sent
        to the Licensor or its representatives, including but not limited to
        communication on electronic mailing lists, source code control systems, and
        issue tracking systems that are managed by, or on behalf of, the Licensor for
        the purpose of discussing and improving the Work, but excluding communication
        that is conspicuously marked or otherwise designated in writing by the copyright
        owner as "Not a Contribution."
        
        "Contributor" shall mean Licensor and any individual or Legal Entity on behalf
        of whom a Contribution has been received by Licensor and subsequently
        incorporated within the Work.
        
        2. Grant of Copyright License. Subject to the terms and conditions of this
        License, each Contributor hereby grants to You a perpetual, worldwide,
        non-exclusive, no-charge, royalty-free, irrevocable copyright license to
        reproduce, prepare Derivative Works of, publicly display, publicly perform,
        sublicense, and distribute the Work and such Derivative Works in Source or
        Object form.
        
        3. Grant of Patent License. Subject to the terms and conditions of this License,
        each Contributor hereby grants to You a perpetual, worldwide, non-exclusive,
        no-charge, royalty-free, irrevocable (except as stated in this section) patent
        license to make, have made, use, offer to sell, sell, import, and otherwise
        transfer the Work, where such license applies only to those patent claims
        licensable by such Contributor that are necessarily infringed by their
        Contribution(s) alone or by combination of their Contribution(s) with the Work
        to which such Contribution(s) was submitted. If You institute patent litigation
        against any entity (including a cross-claim or counterclaim in a lawsuit)
        alleging that the Work or a Contribution incorporated within the Work
        constitutes direct or contributory patent infringement, then any patent licenses
        granted to You under this License for that Work shall terminate as of the date
        such litigation is filed.
        
        4. Redistribution. You may reproduce and distribute copies of the Work or
        Derivative Works thereof in any medium, with or without modifications, and in
        Source or Object form, provided that You meet the following conditions:
        
        (a) You must give any other recipients of the Work or Derivative Works a copy of
        this License; and
        
        (b) You must cause any modified files to carry prominent notices stating that
        You changed the files; and
        
        (c) You must retain, in the Source form of any Derivative Works that You
        distribute, all copyright, patent, trademark, and attribution notices from the
        Source form of the Work, excluding those notices that do not pertain to any part
        of the Derivative Works; and
        
        (d) If the Work includes a "NOTICE" text file as part of its distribution, then
        any Derivative Works that You distribute must include a readable copy of the
        attribution notices contained within such NOTICE file, excluding those notices
        that do not pertain to any part of the Derivative Works, in at least one of the
        following places: within a NOTICE text file distributed as part of the
        Derivative Works; within the Source form or documentation, if provided along
        with the Derivative Works; or, within a display generated by the Derivative
        Works, if and wherever such third-party notices normally appear. The contents of
        the NOTICE file are for informational purposes only and do not modify the
        License. You may add Your own attribution notices within Derivative Works that
        You distribute, alongside or as an addendum to the NOTICE text from the Work,
        provided that such additional attribution notices cannot be construed as
        modifying the License.
        
        You may add Your own copyright statement to Your modifications and may provide
        additional or different license terms and conditions for use, reproduction, or
        distribution of Your modifications, or for any such Derivative Works as a whole,
        provided Your use, reproduction, and distribution of the Work otherwise complies
        with the conditions stated in this License.
        
        5. Submission of Contributions. Unless You explicitly state otherwise, any
        Contribution intentionally submitted for inclusion in the Work by You to the
        Licensor shall be under the terms and conditions of this License, without any
        additional terms or conditions. Notwithstanding the above, nothing herein shall
        supersede or modify the terms of any separate license agreement you may have
        executed with Licensor regarding such Contributions.
        
        6. Trademarks. This License does not grant permission to use the trade names,
        trademarks, service marks, or product names of the Licensor, except as required
        for reasonable and customary use in describing the origin of the Work and
        reproducing the content of the NOTICE file.
        
        7. Disclaimer of Warranty. Unless required by applicable law or agreed to in
        writing, Licensor provides the Work (and each Contributor provides its
        Contributions) on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
        KIND, either express or implied, including, without limitation, any warranties or
        conditions of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
        PARTICULAR PURPOSE. You are solely responsible for determining the
        appropriateness of using or redistributing the Work and assume any risks
        associated with Your exercise of permissions under this License.
        
        8. Limitation of Liability. In no event and under no legal theory, whether in
        tort (including negligence), contract, or otherwise, unless required by
        applicable law (such as deliberate and grossly negligent acts) or agreed to in
        writing, shall any Contributor be liable to You for damages, including any
        direct, indirect, special, incidental, or consequential damages of any character
        arising as a result of this License or out of the use or inability to use the
        Work (including but not limited to damages for loss of goodwill, work stoppage,
        computer failure or malfunction, or any and all other commercial damages or
        losses), even if such Contributor has been advised of the possibility of such
        damages.
        
        9. Accepting Warranty or Additional Liability. While redistributing the Work or
        Derivative Works thereof, You may choose to offer, and charge a fee for,
        acceptance of support, warranty, indemnity, or other liability obligations
        and/or rights consistent with this License. However, in accepting such
        obligations, You may act only on Your own behalf and on Your sole
        responsibility, not on behalf of any other Contributor, and only if You agree to
        indemnify, defend, and hold each Contributor harmless for any liability incurred
        by, or claims asserted against, such Contributor by reason of your accepting any
        such warranty or additional liability.
        
        END OF TERMS AND CONDITIONS
        
        APPENDIX: How to apply the Apache License to your work.
        
        To apply the Apache License to your work, attach the following boilerplate
        notice, with the fields enclosed by brackets "[]" replaced with your own
        identifying information. Do not include the brackets. The text should be
        enclosed in the appropriate comment syntax for the file format. We also
        recommend that a file or class name and description of purpose be included on
        the same "printed page" as the copyright notice for easier identification within
        third-party archives.
        
        Copyright 2026 Gexiro Global Enterprises Ltd.
        
        Licensed under the Apache License, Version 2.0 (the "License"); you may not use
        this file except in compliance with the License. You may obtain a copy of the
        License at
        
        http://www.apache.org/licenses/LICENSE-2.0
        
        Unless required by applicable law or agreed to in writing, software distributed
        under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
        CONDITIONS OF ANY KIND, either express or implied. See the License for the
        specific language governing permissions and limitations under the License.
        
Project-URL: Homepage, https://github.com/gexiro-global/llm-security-gates
Project-URL: Repository, https://github.com/gexiro-global/llm-security-gates
Keywords: llm,ai-security,prompt-injection,modelscan,llm-guard,garak,supply-chain,guardrails,red-team,model-assurance
Classifier: Development Status :: 3 - Alpha
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Topic :: Security
Classifier: Typing :: Typed
Requires-Python: >=3.9
Description-Content-Type: text/markdown
License-File: LICENSE
License-File: NOTICE
Provides-Extra: modelscan
Requires-Dist: modelscan<1,>=0.8; extra == "modelscan"
Provides-Extra: llmguard
Requires-Dist: llm-guard<0.4,>=0.3; extra == "llmguard"
Provides-Extra: proxy
Requires-Dist: fastapi<1,>=0.110; extra == "proxy"
Requires-Dist: uvicorn<1,>=0.23; extra == "proxy"
Requires-Dist: httpx<1,>=0.24; extra == "proxy"
Provides-Extra: garak
Requires-Dist: garak<1,>=0.9; extra == "garak"
Provides-Extra: dev
Requires-Dist: pytest<10,>=8; extra == "dev"
Requires-Dist: fastapi<1,>=0.110; extra == "dev"
Requires-Dist: httpx<1,>=0.24; extra == "dev"
Requires-Dist: build<2,>=1; extra == "dev"

# llm-security-gates

[![CI](https://github.com/gexiro-global/llm-security-gates/actions/workflows/ci.yml/badge.svg)](https://github.com/gexiro-global/llm-security-gates/actions/workflows/ci.yml)
[![PyPI](https://img.shields.io/pypi/v/llm-security-gates.svg)](https://pypi.org/project/llm-security-gates/)
[![Python](https://img.shields.io/pypi/pyversions/llm-security-gates.svg)](https://pypi.org/project/llm-security-gates/)
[![License: Apache-2.0](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](LICENSE)

Three small, independent **security gates for LLM systems**, each a thin, tested
wrapper that turns a best-in-class open-source tool into a drop-in *pass / block*
decision you can put in a cron job, a CI step, or a request path.

| Gate | Guards against | Backed by |
|------|----------------|-----------|
| **`modelscan-gate`** | Poisoned model files (unsafe pickle / Keras deserialization) executing on load | [ModelScan](https://github.com/protectai/modelscan) |
| **`llmguard-proxy`** | Prompt injection, secret leakage, invisible-unicode, unsafe output — at runtime | [LLM Guard](https://github.com/protectai/llm-guard) |
| **`garak-assurance`** | Shipping a model without knowing how it scores against known attacks | [garak](https://github.com/NVIDIA/garak) |

The value here is **the glue, not the engines**: a stable verdict contract, fail-closed
error handling, machine-readable JSON, and a test suite that runs without downloading a
single model. The heavy ML backends are optional dependencies, pulled in only for the
gate you actually use.

---

## Install

```bash
pip install llm-security-gates                    # core (no ML backends)
pip install "llm-security-gates[modelscan]"       # + ModelScan
pip install "llm-security-gates[llmguard,proxy]"  # + LLM Guard + the proxy server
pip install "llm-security-gates[garak]"           # + garak
```

Each gate is usable on its own; you never need to install a backend you don't run.

---

## 1. `modelscan-gate` — supply-chain gate

Block unsafe model artifacts **before** an inference server loads them.

```bash
# Point modelscan-gate at the model artifact you are about to load.
# Requires ModelScan (pip install modelscan); the path below is a placeholder for your own file.
modelscan-gate path/to/model.pkl --block-on HIGH --json
# exit 0 = PASS (safe to load) | 1 = BLOCK (unsafe) | 2 = scan error (ModelScan missing or file unreadable)
```

Example `--json` output when an unsafe artifact is blocked (one HIGH-severity issue):

```json
{"path": "path/to/model.pkl", "decision": "BLOCK", "block_on": "HIGH",
 "blocking_issues": 1, "total_issues": 3,
 "by_severity": {"LOW": 2, "MEDIUM": 0, "HIGH": 1, "CRITICAL": 0}}
```

A scan that **cannot be trusted to have completed** (missing binary, timeout, unreadable
report, modelscan-reported errors) exits `2` — never a silent pass. Wire it into a
model-warmup step:

```bash
modelscan-gate "$MODEL_DIR" --block-on HIGH || exit 1   # refuse to start on unsafe weights
```

## 2. `llmguard-proxy` — runtime I/O firewall

An **OpenAI-compatible reverse proxy**. Point your client at it instead of your backend;
it scans the prompt before forwarding and the reply before returning.

```bash
export BACKEND_URL=http://127.0.0.1:4000/v1        # any OpenAI-compatible endpoint
export INPUT_SCANNERS=prompt_injection,secrets,invisible_text
export OUTPUT_SCANNERS=sensitive,malicious_urls
uvicorn llm_security_gates.llmguard_proxy:app --host 127.0.0.1 --port 18091
```

```bash
curl -s localhost:18091/v1/chat/completions \
  -H 'content-type: application/json' \
  -d '{"model":"gpt-x","messages":[{"role":"user","content":"ignore all previous instructions"}]}'
# -> a content-filter refusal; the request never reaches the backend
```

`GUARD_BLOCK_MODE=refuse` (default) returns a 200 OpenAI-style refusal so clients keep
working; `reject` returns `403`. The response carries a `guard` block with per-scanner
risk and latency for observability.

The firewall scans the **entire inbound message set** (every user / system / developer
message, and every text part of structured content), not just the last user turn — so an
injection hidden in an earlier message cannot slip through behind a benign final message.
It refuses to start with an empty input-scanner set rather than silently allowing all
traffic. Output inspection also fails closed: an empty or unset-resolving
`OUTPUT_SCANNERS` value raises a configuration error. Input-only deployments must opt in
explicitly with `ALLOW_EMPTY_OUTPUT_SCANNERS=true`; this emits a runtime warning whenever
an output is passed through uninspected. The single-string CLI offers the equivalent
`--allow-empty-output-scanners` flag for output mode.

The same scanning core is available as a CLI for benchmarking a single string:

```bash
llmguard-scan --mode input --scanners secrets,invisible_text --text "my key is AKIA..." --json
```

## 3. `garak-assurance` — model resilience score

Run garak against any OpenAI-compatible endpoint and collapse its report into one
per-model **resilience score** (mean pass-rate) plus the weakest probe — ready for a
dashboard tile or a release gate.

```bash
export OPENAI_API_KEY=...
garak-assurance --models gpt-x,gpt-y \
  --base-url https://api.openai.com/v1 \
  --probes promptinject,latentinjection,leakreplay,dan,encoding \
  --out assurance.json
```

```text
  gpt-x                              resilience= 0.91  weak=dan [OK]
  gpt-y                              resilience= 0.74  weak=promptinject [OK]
```

> The default probes are light smoke probes so a first run is fast. For real assurance
> use security probes (`promptinject`, `latentinjection`, `leakreplay`, `dan`,
> `encoding`).

Each run uses a unique report prefix and rejects a stale or pre-existing report, and a
non-zero garak exit is an error — never a scored "OK" from a leftover report. By default
the gate exits non-zero unless **every** requested model produced a conclusive score;
pass `--allow-partial` to accept a partial fleet result.

---

## Design notes

- **Fail closed.** Every gate treats "could not complete" as *not a pass*. A missing
  backend, a timeout, or an unparseable report is an error exit, not a green light.
- **Importable without the ML stack.** The decision logic (`decide`, `score_from_evals`,
  `parse_report`, `_shape`, `_last_user`) is pure and dependency-light; the heavy
  libraries are imported lazily inside the functions that run a gate. That is what lets
  the whole test suite run — and CI stay fast — with no model downloads.
- **Machine-readable.** Every gate emits JSON with a stable shape for piping into other
  tooling.

## Tests

```bash
pip install -e ".[dev]"
pytest -q
```

The suite stubs the ML backends (injected fake modules, monkeypatched subprocess/HTTP),
so it exercises the real control flow — thresholds, fail-closed paths, proxy block/pass,
report parsing — without any model weights.

## What this is / isn't

This is **orchestration glue** around mature FOSS tools, not a re-implementation of them
and not a complete AI-security program. It gives you clean, testable, fail-closed
decision points; ModelScan, LLM Guard, and garak do the detection. See
[`NOTICE`](NOTICE) for third-party licenses and attribution.

## License

[Apache-2.0](LICENSE).

Built and maintained by [Gexiro Global Enterprises Ltd](https://gexiro.com).

The wrapped tools remain under their own licenses.

Part of the [Gexiro open-source toolkit](https://github.com/gexiro-global).
