| Version | Supported |
|---|---|
| 0.1.x | Yes |
| < 0.1 | No |
This repository is an early research codebase. Security support focuses on preventing credential leakage, unsafe recovery actions, and malicious benchmark or example code.
Please do not open a public GitHub issue for security vulnerabilities.
Instead:
- Email the maintainer via the address listed on the askmy-stack GitHub profile, or
- Use GitHub's private vulnerability reporting for this repository (Security → Report a vulnerability), if enabled.
Include:
- a description of the issue and impact
- steps to reproduce
- affected files / versions if known
You should receive an acknowledgement within 7 days. We will coordinate a fix and disclosure timeline after triage.
- Benchmark failure injectors must not execute arbitrary remote code by default.
- Examples must not ship live credentials or private API keys.
- Recovery actions that mutate external state should be sandboxed or clearly marked as unsafe for production use.
- LLM-based judges and detectors should avoid sending secrets in prompts.