prompt
FreeDesign agent cooperation with clear rules and anti-collusion checks
FreeFree tier
Inputs: textOutputs: text
About prompt
A system prompt for designing cooperative multi-agent systems, part of the ai-boost/awesome-prompts collection. It guides the creation of a structured framework for agent collaboration, including shared objectives, local responsibilities, cooperation triggers, and anti-collusion checks. The prompt emphasizes distinguishing genuine cooperation from blind agreement and outputs eight specific sections: Task Goal, Shared Objective, Agent Roles, Cooperation Rules, Disagreement/Challenge Rules, Anti-Herding Controls, Evaluation Signals, and Main Tradeoff.
Key Features
Designs shared objective and local responsibilities for agents
Defines cooperation triggers and escalation rules
Includes anti-collusion checks and independent validation requirements
Outputs structured sections: Task Goal, Agent Roles, Cooperation Rules, etc.
Emphasizes distinguishing cooperation from blind agreement
Cites academic sources (arXiv, LangMARL, Agent Q-Mix)
Pros & Cons
Pros
- Encourages thoughtful design of agent interactions with explicit tradeoffs
- Explicitly prevents herd errors and blind agreement through anti-collusion checks
- Provides a clear, structured output format for consistent results
- Open source and freely available on GitHub
Cons
- Niche use case for multi-agent system designers only
- Requires domain knowledge in agent cooperation and game theory
- Output is a structured text plan, not executable code or a ready-to-use system
- No built-in validation or testing of the generated cooperation design
Best For
Designing multi-agent AI systems for cooperative tasksCreating agent interaction frameworks in games or simulationsStructuring agent teams with clear role definitions and decision rightsTeaching or prototyping cooperative multi-agent architectures
FAQ
What is the purpose of this prompt?
To guide the design of incentives, roles, and coordination rules for multiple AI agents, ensuring cooperation when beneficial and exposing when competition or independence is healthier.
What sections must the output include?
Task Goal, Shared Objective, Agent Roles, Cooperation Rules, Disagreement/Challenge Rules, Anti-Herding Controls, Evaluation Signals, and Main Tradeoff.
How does this prompt handle the risk of agents blindly agreeing?
It requires explicit anti-collusion checks, independent validation requirements, and mechanisms to surface dissent, ensuring cooperation is not confused with blind agreement.