Optional OSS share: Agent Memory Guard for LangChain-style agent memory

[vgudur-dev]

Hi — optional heads-up for agents that keep chat/checkpoint memory.

OWASP Agent Memory Guard is an OSS project aimed at reducing agent-memory poisoning / injection risk in LangChain-style agent loops. There’s a small LangChain-oriented package and a short clinic gist if useful:

• https://github.com/OWASP/www-project-agent-memory-guard
• https://pypi.org/project/langchain-agent-memory-guard/
• https://gist.github.com/vgudur-dev/ead8817d2f4df08c04b808cc5b53eb06
• How-to Discussion: How-to: Protect LangChain agents with langchain-agent-memory-guard (5 minutes) OWASP/www-project-agent-memory-guard#116
  Only sharing in case it’s relevant to SurfSenses create_agent / Deep Agents stack — happy to answer technical questions; no ask beyond a look if interested.