Multi-VPS orchestration
I’m designing a multi-agent system distributed across several VPS instances and wanted feedback on whether this architecture makes sense or if there are obvious failure points.
1. Research Agent (Standalone Instance)
Dedicated to structured research tasks such as:
- Market/competitor analysis
- Technical research
- Ongoing recurring searches
This agent maintains a shared research knowledge base used by all other agents.
Basic browsing capability still exists in other agents when needed.
2. Developer Agent (Separate Instance)
- Handles engineering and technical execution tasks.
- Runs independently from research workloads.
3. Orchestration Layer (Separate Server)
Acts as the central control layer:
Responsibilities:
- Validate outputs from other agents
- Coordinate tasks across agents
- Improve or modify agent configurations/files
- Maintain system stability
Additional functions:
- Automatically provision new VPS instances for new agent deployments
My main idea is to prevent the orchestration layer from getting prompt-injected or ever going down. I want to have separation between the agents so that certain agents won’t ever go down in case there’s an issue with another agent’s configuration.
My main concerns are whether agent-to-agent communication and syncing will be efficient, and whether there is anything already built-in that can handle that. I’m also not sure whether this approach is the best one. Another option could be having Docker containers per agent, but if there is a misconfiguration in the gateway, all agents could stop working.
1. Research Agent (Standalone Instance)
Dedicated to structured research tasks such as:
- Market/competitor analysis
- Technical research
- Ongoing recurring searches
This agent maintains a shared research knowledge base used by all other agents.
Basic browsing capability still exists in other agents when needed.
2. Developer Agent (Separate Instance)
- Handles engineering and technical execution tasks.
- Runs independently from research workloads.
3. Orchestration Layer (Separate Server)
Acts as the central control layer:
Responsibilities:
- Validate outputs from other agents
- Coordinate tasks across agents
- Improve or modify agent configurations/files
- Maintain system stability
Additional functions:
- Automatically provision new VPS instances for new agent deployments
My main idea is to prevent the orchestration layer from getting prompt-injected or ever going down. I want to have separation between the agents so that certain agents won’t ever go down in case there’s an issue with another agent’s configuration.
My main concerns are whether agent-to-agent communication and syncing will be efficient, and whether there is anything already built-in that can handle that. I’m also not sure whether this approach is the best one. Another option could be having Docker containers per agent, but if there is a misconfiguration in the gateway, all agents could stop working.
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
130,529Members
Resources
Similar Threads
Was this page helpful?
