Pryxor is an open-source runtime security gateway that sits between AI agents and external systems, evaluating and controlling each action before execution. It enforces explicit policies to approve safe actions, block forbidden ones, or pause risky actions for human review, addressing the gap between identity authorization and action-level security for autonomous agents.
llm-gauze is an HTTP gateway that sits between clients and local LLMs to handle transient errors, retries, hung models, malformed responses, and context overflows. It runs as a Docker container with an OpenAI-compatible API, configured via environment variables to point at a local LLM endpoint, and logs all exchanges for debugging and remediation.