| Version | Supported |
|---|---|
| 0.1.x | ✅ |
If you discover a security vulnerability in this project, please report it responsibly.
- Do NOT open a public GitHub issue for security vulnerabilities
- Email the maintainers directly or use GitHub's private vulnerability reporting feature
- Include:
- Description of the vulnerability
- Steps to reproduce
- Potential impact
- Any suggested fixes (optional)
- Acknowledgment of your report within 48 hours
- Regular updates on the progress of the fix
- Credit in the security advisory (unless you prefer to remain anonymous)
This demo is designed to run entirely on your local machine:
- No cloud API keys required — All inference happens via Foundry Local
- Data stays on-device — Documents in the
data/folder are never sent to external services - No network calls during inference — After initial model download, the app works offline
- Web UI binds to localhost — The Flask web server runs on
127.0.0.1by default and is not exposed to the network
- Models are downloaded exclusively from Microsoft's official Foundry Local catalog
- Verify model provenance using
foundry model listto see official model IDs and sources - Models are cached locally in the Foundry Local cache directory
- Do not load models from untrusted sources or manually override model paths with unverified files
- You can use trusted Hugging Face models using custom models see https://techcommunity.microsoft.com/blog/educatordeveloperblog/how-to-use-custom-models-with-foundry-local-a-beginners-guide/4428857
- The web UI (
web.py) is intended for local development only — do not deploy it to production it current has NO authentication, CSRF protection, and rate limiting - The
/api/runendpoint executes agent workflows — ensure it is not exposed to untrusted networks - Input sanitisation is applied but the application is not hardened for adversarial use this is a demo only.
- The
.envfile may contain configuration settings - Never commit
.envfiles with sensitive data to version control - Use
.env.exampleas a template
- Keep dependencies updated:
pip install --upgrade -r requirements.txt - Review
pyproject.tomlfor pinned versions - Run
pip auditperiodically to check for known vulnerabilities
This project depends on:
- foundry-local-sdk — Microsoft Foundry Local SDK
- agent-framework-core — Microsoft Agent Framework
- openai — OpenAI Python SDK (used for the OpenAI-compatible API)
Report vulnerabilities in these dependencies to their respective maintainers.