on-policy
★ 2.1k
vs
tunnel-client
★ 375
on-policy vs tunnel-client
on-policy: This repository implements MAPPO, a multi-agent variant of PPO, widely used in cooperative multi-agent games and research. It provides robust implementations for various multi-agent environments like StarCraft II, Hanabi, and Google Research Football, along with detailed training scripts and hyperparameter guidance.; tunnel-client: `tunnel-client` is a customer-run agent that securely connects private or localhost Model Context Protocol (MCP) servers to OpenAI products like ChatGPT and Codex through an OpenAI-hosted tunnel. It enables secure access for AI models without exposing the MCP server to the public internet, simplifying deployments in restricted network environments.
01
TL;DR
Choose on-policy if…Research and experimentation in cooperative multi-agent reinforcement learning
Choose tunnel-client if…Connecting a private or localhost MCP server to OpenAI-hosted products like ChatGPT when public internet access is restricted.
02
Side-by-Side Comparison
03
Features
on-policy01Implementation of MAPPO (Multi-Agent PPO)
02Support for diverse multi-agent environments (e.g., StarCraft II, Hanabi)
03Ready-to-use training scripts for various scenarios
04Detailed hyperparameter guidance and updated results
05Default support for shared policy among agents
tunnel-client01Securely connects private or localhost MCP servers to OpenAI products like ChatGPT and Codex.
02Operates as a daemon with health checks, metrics, and an administrative UI for operational visibility.
03Supports flexible MCP server integration via Streamable HTTP, stdio, or in-memory transports.
04Offers an embeddable Go SDK for direct integration into Go applications.
04
Use Cases
on-policy↳Research and experimentation in cooperative multi-agent reinforcement learning
↳Benchmarking and evaluating PPO's effectiveness in MARL scenarios
↳Training AI agents for popular multi-agent games like StarCraft II and Hanabi
tunnel-client↳Connecting a private or localhost MCP server to OpenAI-hosted products like ChatGPT when public internet access is restricted.
↳Deploying an operator-visible daemon for an MCP server before connecting it to external APIs, ensuring readiness and observability.
↳Integrating directly into a Go application using the Go SDK to bridge an in-memory MCP server with OpenAI tunnels.
05
Best For
on-policyTrendingReinforcement LearningMulti-Agent AI
tunnel-clientEssentialHidden Gem
FAQ
FAQ
What is the difference between on-policy and tunnel-client?
Both on-policy and tunnel-client are in the LLM Infra category. on-policy has 2.1k stars, while tunnel-client has 375 stars.
Which is better, on-policy or tunnel-client?
The best choice depends on your use case. Choose on-policy if Research and experimentation in cooperative multi-agent reinforcement learning, and tunnel-client if Connecting a private or localhost MCP server to OpenAI-hosted products like ChatGPT when public internet access is restricted..
Is on-policy free or open source?
Yes, on-policy is open source on GitHub (MIT).
Is tunnel-client free or open source?
Yes, tunnel-client is open source on GitHub (Apache-2.0).
→