githuganswered from live GitHub & Hugging Face data
Best open-source LLM gateways for running multiple coding agents simultaneously?
To run multiple coding agents simultaneously through Open Source LLM (Large Language Model) gateways, you have an attractive set of options, each with distinct capabilities and current activity. Here’s a rundown:
The Contenders
ASCIT31/Dark-Moon
Stars: 1,005
Language: Python
License: GPL-3.0
Maintenance: ✅ Actively maintained (last pushed 5 days ago)
Overview: An open-source AI penetration testing framework integrating around 50 specialist agents that operate on a local LLM through a Privacy Gateway. It seamlessly connects with platforms like GitHub and Jenkins.apache/casbin-gateway
Stars: 641
Language: Go
License: Apache-2.0
Maintenance: ✅ Actively maintained (last pushed 2 days ago)
Overview: This gateway monitors various coding agents, providing one endpoint for multiple LLM API vendors. It offers authentication grading and detailed usage records, making it an excellent choice for managing agent interactions at scale.esinecan/skynet-agent
Stars: 135
Language: TypeScript
License: MIT
Maintenance: ⚠️ Maintained (last pushed 165 days ago)
Overview: This agent grants access to various LLMs while implementing dual-layer memory architecture. It supports automatic and conscious memory operations, allowing for advanced agent behavior but hasn’t been updated as recently.wink-run/tokenbank
Stars: 104
Language: JavaScript
License: Apache-2.0
Maintenance: ✅ Actively maintained (last pushed 5 days ago)
Overview: A local LLM gateway designed to optimize token usage by routing requests intelligently to multiple model providers. It offers an easy setup for various agents, with tracking of token expenditures.elliot35/deterministic-agent-control-protocol
Stars: 82
Language: TypeScript
License: MIT
Maintenance: ⚠️ Slowing (last pushed 236 days ago)
Overview: A governance layer for AI agents, enabling session-aware control and auditing. While it supports multiple agents, its slowdown in updates suggests a cautious approach.
Comparative Table
| Project | Stars | Language | License | Maintenance | Last Push | Description |
|---|---|---|---|---|---|---|
| ASCIT31/Dark-Moon | 1,005 | Python | GPL-3.0 | ✅ Active | 5 days ago | AI penetration testing with 50+ specialist agents. |
| apache/casbin-gateway | 641 | Go | Apache-2.0 | ✅ Active | 2 days ago | Monitors multiple coding agents with a focus on API interactions. |
| esinecan/skynet-agent | 135 | TypeScript | MIT | ⚠️ Maintained | 165 days ago | Supports most LLMs with dual-layer memory operations. |
| wink-run/tokenbank | 104 | JavaScript | Apache-2.0 | ✅ Active | 5 days ago | Local LLM gateway for token optimization across multiple agents. |
| elliot35/deterministic-agent-control-protocol | 82 | TypeScript | MIT | ⚠️ Slowing | 236 days ago | Governance protocol for AI agents with auditing features. |
Verdict
For a best-in-class experience: ASCIT31/Dark-Moon stands out with robust features and current activity. It's tailored for security testing with extensive agent integration.
For API monitoring and control: Choose apache/casbin-gateway to keep tight oversight on multiple coding agents and ensure compliance through specialized monitoring.
Opt for newer, flexible setups: If you seek fresh routes with token management, wink-run/tokenbank is your go-to, blending economy and usability well in recent development.
Recommendation
Choose Dark-Moon if you focus on security and automation; select Casbin-gateway for managing diverse agent requests; Tokenbank is ideal if optimizing resource usage is your priority.