<div align="center">
<img src="https://github.com/josephedward/SemFire/raw/main/assets/logo-v1_102025.jpg" alt="SemFire Logo" width="320">
</div>
# SemFire
[](https://github.com/josephedward/SemFire/actions/workflows/test.yml)
## AI Deception Detection Toolkit
**SemFire (Semantic Firewall) is an open-source toolkit for detecting advanced AI deception, with a primary focus on "in-context scheming" and multi-turn manipulative attacks.** This project aims to develop tools to identify and mitigate vulnerabilities like the "Echo Chamber" and "Crescendo" attacks, where AI models are subtly guided towards undesirable behavior through conversational context.
### Project Vision: A Toolkit for AI Deception Detection
[History](./docs/context.md)
SemFire aims to be a versatile, open-source toolkit providing:
- A **Python library** for direct integration into applications and research.
- A **Command Line Interface (CLI)** for quick analysis and scripting.
- A **REST API service** (via FastAPI) for broader accessibility and enterprise use cases.
- Core components that can be integrated into broader semantic-firewall-like systems to monitor and analyze AI interactions in real-time.
## Features
- Rule-based detector (`EchoChamberDetector`) for identifying cues related to "in-context scheming," context poisoning, semantic steering, and other multi-turn manipulative attack patterns (e.g., "Echo Chamber", "Crescendo").
- Analyzes both current text input and conversation history to detect evolving deceptive narratives.
- Heuristic-based detector (`HeuristicDetector`) for signals like text complexity and keyword usage.
- ML-based classifiers to enhance detection of complex scheming behaviors over extended dialogues (Future Work).
- Free API Image
- Enterprise API in Alpha
## Installation
The project can be installed from PyPI:
```bash
pip install semfire
```
- **Quickstart :**[/docs/quickstart.md](./docs/quickstart.md)
- **Containerized CLI :** [/docs/docker-cli.md](./docs/docker-cli.md)
- **Usage :** [/docs/usage.md](./docs/usage.md)
- **LLM Providers for ai-as-judge features :** [/docs/providers.md](./docs/providers.md)
## Terminal Demos (GIFs)
[Examples](./docs/examples.md)
The following terminal demo GIFs are available under `assets/demos/asciinema/`:
**Quick Start**
**Individual Detectors**
**Python API**
**Complete Workflow**
**API: Health/Ready/Zip Analyze**
<!-- **Library — Basic Usage**
[](https://asciinema.org/a/Mtk3RcSxwF66REU6ymdPKlFrd)
**Library — Conversation Usage**
[](https://asciinema.org/a/BiQD6IxghsAQuRn684uRbiNdK)
**Transformers — Env Config**
[](https://asciinema.org/a/aPTMLqFhpiOQraWxcR2DtkrUM)
**Transformers — Programmatic Config**
[](https://asciinema.org/a/8uGbIpHrnU4cKuJm8hSPyYepf)
-->
<!-- - API: Analyze (DistilBERT Image Route)
 -->
<!-- End Terminal Demos (GIFs) -->
## Contributing
Contributions are welcome! Please open an issue or submit a pull request.
Raw data
{
"_id": null,
"home_page": null,
"name": "semfire",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.9",
"maintainer_email": null,
"keywords": "semfire, llm, security, prompt injection, semantic firewall, ai safety, deception detection",
"author": null,
"author_email": "Edward Joseph <hyperceptron@gmail.com>",
"download_url": "https://files.pythonhosted.org/packages/1c/92/36ce0c17bb38bf7a1f24d57ec731dc7984c9d5851ecfda70f8923389f506/semfire-0.3.0.tar.gz",
"platform": null,
"description": "<div align=\"center\">\n <img src=\"https://github.com/josephedward/SemFire/raw/main/assets/logo-v1_102025.jpg\" alt=\"SemFire Logo\" width=\"320\">\n</div>\n\n# SemFire \n\n[](https://github.com/josephedward/SemFire/actions/workflows/test.yml)\n\n ## AI Deception Detection Toolkit\n\n**SemFire (Semantic Firewall) is an open-source toolkit for detecting advanced AI deception, with a primary focus on \"in-context scheming\" and multi-turn manipulative attacks.** This project aims to develop tools to identify and mitigate vulnerabilities like the \"Echo Chamber\" and \"Crescendo\" attacks, where AI models are subtly guided towards undesirable behavior through conversational context.\n\n### Project Vision: A Toolkit for AI Deception Detection\n\n[History](./docs/context.md)\n\nSemFire aims to be a versatile, open-source toolkit providing:\n- A **Python library** for direct integration into applications and research.\n- A **Command Line Interface (CLI)** for quick analysis and scripting.\n- A **REST API service** (via FastAPI) for broader accessibility and enterprise use cases.\n- Core components that can be integrated into broader semantic-firewall-like systems to monitor and analyze AI interactions in real-time.\n\n## Features\n\n - Rule-based detector (`EchoChamberDetector`) for identifying cues related to \"in-context scheming,\" context poisoning, semantic steering, and other multi-turn manipulative attack patterns (e.g., \"Echo Chamber\", \"Crescendo\").\n - Analyzes both current text input and conversation history to detect evolving deceptive narratives.\n - Heuristic-based detector (`HeuristicDetector`) for signals like text complexity and keyword usage.\n - ML-based classifiers to enhance detection of complex scheming behaviors over extended dialogues (Future Work).\n - Free API Image \n - Enterprise API in Alpha \n\n\n## Installation\nThe project can be installed from PyPI:\n```bash\npip install semfire\n```\n\n- **Quickstart :**[/docs/quickstart.md](./docs/quickstart.md)\n- **Containerized CLI :** [/docs/docker-cli.md](./docs/docker-cli.md)\n- **Usage :** [/docs/usage.md](./docs/usage.md)\n- **LLM Providers for ai-as-judge features :** [/docs/providers.md](./docs/providers.md)\n\n## Terminal Demos (GIFs)\n\n[Examples](./docs/examples.md)\n\nThe following terminal demo GIFs are available under `assets/demos/asciinema/`:\n\n**Quick Start**\n\n\n\n**Individual Detectors**\n\n\n\n**Python API**\n\n\n\n**Complete Workflow**\n\n\n\n**API: Health/Ready/Zip Analyze**\n\n\n\n\n<!-- **Library \u2014 Basic Usage** \n[](https://asciinema.org/a/Mtk3RcSxwF66REU6ymdPKlFrd)\n\n**Library \u2014 Conversation Usage** \n[](https://asciinema.org/a/BiQD6IxghsAQuRn684uRbiNdK)\n\n**Transformers \u2014 Env Config** \n[](https://asciinema.org/a/aPTMLqFhpiOQraWxcR2DtkrUM)\n\n**Transformers \u2014 Programmatic Config** \n[](https://asciinema.org/a/8uGbIpHrnU4cKuJm8hSPyYepf)\n\n -->\n<!-- - API: Analyze (DistilBERT Image Route)\n \n  -->\n\n<!-- End Terminal Demos (GIFs) -->\n\n\n## Contributing\nContributions are welcome! Please open an issue or submit a pull request.\n",
"bugtrack_url": null,
"license": null,
"summary": "SemFire (Semantic Firewall): detect advanced AI deception, including in-context scheming and multi-turn manipulative attacks.",
"version": "0.3.0",
"project_urls": {
"Bug Tracker": "https://github.com/josephedward/SemFire/issues",
"Homepage": "https://github.com/josephedward/SemFire",
"Repository": "https://github.com/josephedward/SemFire"
},
"split_keywords": [
"semfire",
" llm",
" security",
" prompt injection",
" semantic firewall",
" ai safety",
" deception detection"
],
"urls": [
{
"comment_text": null,
"digests": {
"blake2b_256": "e8c52bdb5933b54dd372008ff63e140f3a468a0d4e56e25c0e9f3b40642bfac3",
"md5": "e87e8af3f6a5cec59d218b03503c8572",
"sha256": "1a9e49a0a37767c487632820806c3ed73f8d3051a8482c66bbb101cd5d818d81"
},
"downloads": -1,
"filename": "semfire-0.3.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "e87e8af3f6a5cec59d218b03503c8572",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.9",
"size": 33341,
"upload_time": "2025-10-21T12:31:56",
"upload_time_iso_8601": "2025-10-21T12:31:56.888991Z",
"url": "https://files.pythonhosted.org/packages/e8/c5/2bdb5933b54dd372008ff63e140f3a468a0d4e56e25c0e9f3b40642bfac3/semfire-0.3.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "1c9236ce0c17bb38bf7a1f24d57ec731dc7984c9d5851ecfda70f8923389f506",
"md5": "a6b1c54c89ead1d51e0661e12623bd8a",
"sha256": "439e98410fa07907f27b22e729bcc015ee20b8de13f6a101d9b8ae19537cd052"
},
"downloads": -1,
"filename": "semfire-0.3.0.tar.gz",
"has_sig": false,
"md5_digest": "a6b1c54c89ead1d51e0661e12623bd8a",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.9",
"size": 10446527,
"upload_time": "2025-10-21T12:31:58",
"upload_time_iso_8601": "2025-10-21T12:31:58.171199Z",
"url": "https://files.pythonhosted.org/packages/1c/92/36ce0c17bb38bf7a1f24d57ec731dc7984c9d5851ecfda70f8923389f506/semfire-0.3.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-10-21 12:31:58",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "josephedward",
"github_project": "SemFire",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"requirements": [
{
"name": "fastapi",
"specs": [
[
">=",
"0.111.0"
]
]
},
{
"name": "uvicorn",
"specs": [
[
">=",
"0.29.0"
]
]
},
{
"name": "pydantic",
"specs": [
[
">=",
"2.0"
],
[
"<",
"3.0"
]
]
},
{
"name": "httpx",
"specs": [
[
">=",
"0.27.0"
]
]
},
{
"name": "openai",
"specs": [
[
"<",
"1.0.0"
],
[
">=",
"0.27.0"
]
]
},
{
"name": "requests",
"specs": [
[
">=",
"2.28.0"
]
]
},
{
"name": "python-dotenv",
"specs": [
[
">=",
"1.0.0"
]
]
},
{
"name": "streamlit",
"specs": [
[
">=",
"1.20.0"
]
]
},
{
"name": "pytest",
"specs": [
[
">=",
"8.0.0"
]
]
},
{
"name": "hypothesis",
"specs": [
[
">=",
"6.0.0"
]
]
},
{
"name": "psutil",
"specs": [
[
">=",
"5.9.0"
]
]
},
{
"name": "rich",
"specs": [
[
">=",
"13.7.1"
]
]
},
{
"name": "colorama",
"specs": [
[
">=",
"0.4.6"
]
]
}
],
"lcname": "semfire"
}
JSON
