Skip to content

Security: GijungKim/expo-local-llm

Security

SECURITY.md

Security

IMPORTANT

We do not accept AI-generated security reports. If you submit one, it will be ignored.

Threat Model

expo-local-llm runs LLM inference entirely on-device. No user data leaves the device through this module — prompts and responses stay local.

Out of Scope

Category Rationale
LLM output content On-device models may produce inaccurate or inappropriate content; this is inherent to LLMs, not a vulnerability in this module
Apple Intelligence / Gemini Nano behavior Model behavior is controlled by Apple/Google, not this module
MCP server interactions Not supported in this module
App-level data handling How the consuming app stores or transmits LLM responses is outside our scope

Reporting Security Issues

To report a security issue, please use the GitHub Security Advisory "Report a Vulnerability" tab.

We will send a response indicating the next steps in handling your report. After the initial reply, we will keep you informed of progress towards a fix.

If you do not receive an acknowledgement within 7 business days, open a regular issue referencing your report.

There aren't any published security advisories