Skip to content

Security: CinvanaAI/python-agent-foundry-workbench

Security

SECURITY.md

Security and privacy

  • No environment file, credential, database, generated agent, chat, transcript, model state, or private connector is included.
  • Generated state defaults beneath the repository's Storage/Generated Artifacts area and is ignored by Git.
  • Embedded package and workflow source is executable Python. Review it as code before running it.
  • User approval screens are workflow controls, not an operating-system sandbox.
  • The historical runner can create and remove its own contained temporary task directories; it should not be pointed at an untrusted storage root.
  • Chatroom files contain plaintext message bodies.

Use this snapshot in a disposable environment for evaluation. Do not load untrusted packages or agent documents.

There aren't any published security advisories