Skip to content

Security: camplus360/agent-memory-bridge

Security

SECURITY.md

Security Policy

Scope

agent-memory-bridge is a local, single-user bridge. Adapters capture agent events on your machine and send them, over loopback HTTP, to a memory worker that is also running locally (default 127.0.0.1:37701 / 127.0.0.1:8000). The project never ships conversation content to a third-party cloud.

Reporting a vulnerability

Please do not open a public issue for security bugs. Email the maintainer:

camplus360@163.com

Include a description, reproduction steps, affected file/version, and (if possible) a suggested fix. You should receive an initial response within a few days. Thank you for disclosing responsibly.

Things to be aware of

  • Do not expose the worker ports. The claude-mem / mem0 workers bind to loopback by default. Binding them to 0.0.0.0 or a public interface would expose stored conversation data to the network; keep them local or behind authentication and a firewall.
  • Secrets in prompts/tool output. Anything your agent sees — including credentials printed by tools — can be captured and stored in the local memory database. Avoid printing secrets, and review the worker's own data-retention settings. .env files are git-ignored; never commit API keys.
  • api passthrough is restricted to paths beginning with /api/ on the configured local worker, to limit server-side request forgery.
  • Installers and hooks run locally; review install.sh and the adapter you enable before running them, as you would with any script.

Dependency / upstream concerns

This repository bundles no third-party runtime code; it interoperates with the separately installed, Apache-2.0-licensed claude-mem and mem0 programs. Security issues in those backends should be reported to their own projects.

There aren't any published security advisories