# Mirror Reflections

> Research notes and interactive security labs from Mirror Security.

The articles explain methods, evidence, measured results and limitations. Live workflow availability varies; saved results are identified separately.

## Research and labs

- [Research Library | Mirror Reflections](https://mirrorsecurity.io/reflections/)
- [Privacy Must Work Both Ways | Mirror Research](https://mirrorsecurity.io/reflections/research/reasoning-replay/)
- [How FHE protects a complete LLM response | Mirror Research](https://mirrorsecurity.io/reflections/research/encrypted-inference/)
- [Double-Blind Evaluation Without a Shared Plaintext Boundary | Mirror Research](https://mirrorsecurity.io/reflections/research/double-blind-evaluation/)
- [When the Goal Finds Another Route | Mirror Reflections](https://mirrorsecurity.io/reflections/research/goal-driven-agent-security/)
- [Authority Conservation for Multi-Agent Systems | Mirror Reflections](https://mirrorsecurity.io/reflections/research/agent-security-scenarios/)
- [AI Safety Should Learn the Signal | Mirror Reflections](https://mirrorsecurity.io/reflections/research/private-ai-oversight/)
- [Zero Data Retention Is Not Encrypted Inference | Mirror Research](https://mirrorsecurity.io/reflections/research/zdr-vs-encrypted-inference/)
- [Mirror Cyber: encrypted AI for the defender’s workflow | Mirror Reflections](https://mirrorsecurity.io/reflections/research/verified-remediation/)
- [Reasoning Replay Protection | Mirror Reflections](https://mirrorsecurity.io/reflections/reasoning-replay/)
- [ZDR and the Encrypted Agent Lifecycle | Mirror Reflections](https://mirrorsecurity.io/reflections/zdr-vs-encrypted-inference/)
- [Mirror Agent Incident Lab | Mirror Reflections](https://mirrorsecurity.io/reflections/agent-incident-lab/)
- [Agent Security Scenario Lab | Mirror Reflections](https://mirrorsecurity.io/reflections/agent-security-scenarios/)