# Memgar (memgar.com)

Inspects LLM agent memory writes and retrievals with pattern, similarity and provenance checks to block poisoned entries.

- Atlas card: https://inite.ai/en/atlas/memgar.com
- Site: https://memgar.com
- For: developers of LLM agents with persistent memory, guarding against memory poisoning
- Categories: dev
- Use 18/100 · alive 82/100

## Used for

- screen memory writes for injected instructions before storage
- detect tampered memory entries after they were stored
- check retrieved chunks and tool results before the model sees them

## Price

Not read yet.

## For developers

None found.
- Read 2026-10-22

## Your data

- Certifications stated: none
- Read 2026-10-22

---

Source: INITE Atlas, https://inite.ai/en/atlas/memgar.com. JSON: https://inite.ai/api/atlas/memgar.com. MCP: https://inite.ai/api/mcp (atlas_lookup).