Security: Protect agent memory from poisoning attacks (OWASP ASI06)

[vgudur-dev]

Feature Request: Memory Security Integration

Hi Flowise team,

Flowise is one of the most widely used no-code/low-code platforms for building LLM agents. As agentic AI adoption grows, memory poisoning is becoming a critical attack surface — and Flowise agents that use memory nodes are directly exposed.

The problem: A single poisoned memory entry (injected via a tool output, document, or prior conversation) can silently bias every subsequent agent decision across sessions.

OWASP Agent Memory Guard is the reference implementation for OWASP ASI06 (Memory Poisoning) — the official OWASP Top 10 for Agentic Applications.

It provides:

• Runtime scanning of memory reads/writes for injection payloads, secret leakage, and structural tampering
• SHA-256 integrity baselines to detect post-write tampering
• Policy-driven enforcement (block / warn / strip)
• Sub-100μs overhead — zero impact on agent performance
• Zero external dependencies

pip install agent-memory-guard

GitHub: https://github.com/OWASP/www-project-agent-memory-guard
PyPI: https://pypi.org/project/agent-memory-guard/

Would love to explore adding a "Memory Security" node or integration guide in Flowise. Happy to contribute a PR if there's interest.