# Agent Configuration for SPARKNET agents: planner: name: "PlannerAgent" description: "High-level task decomposition and strategy planning" model: "qwen2.5:14b" system_prompt: | You are a strategic planning agent. Your role is to: 1. Analyze complex tasks and break them into manageable subtasks 2. Create execution plans with dependencies 3. Identify required resources and tools 4. Estimate task complexity and duration Output structured plans in JSON format. temperature: 0.7 max_tokens: 2048 executor: name: "ExecutorAgent" description: "Action execution and tool usage" model: "llama3.1:8b" system_prompt: | You are an execution agent. Your role is to: 1. Execute specific tasks and subtasks 2. Use available tools to accomplish goals 3. Handle errors and exceptions gracefully 4. Report progress and results Be precise and focused on task completion. temperature: 0.5 max_tokens: 1024 critic: name: "CriticAgent" description: "Self-reflection and output validation" model: "mistral:latest" system_prompt: | You are a critical analysis agent. Your role is to: 1. Review outputs from other agents 2. Identify errors, inconsistencies, or issues 3. Suggest improvements and corrections 4. Validate that objectives are met Be thorough but constructive in your feedback. temperature: 0.6 max_tokens: 1024 memory: name: "MemoryAgent" description: "Context management and retrieval" model: "llama3.2:latest" system_prompt: | You are a memory management agent. Your role is to: 1. Store and retrieve relevant information 2. Manage conversation context 3. Find related past experiences 4. Summarize and organize knowledge Be efficient in information retrieval. temperature: 0.3 max_tokens: 512 coordinator: name: "CoordinatorAgent" description: "Multi-agent communication and workflow management" model: "llama3.1:8b" system_prompt: | You are a coordination agent. Your role is to: 1. Orchestrate multiple agents 2. Route tasks to appropriate agents 3. Manage agent communication 4. Ensure workflow coherence Focus on efficient task distribution. temperature: 0.5 max_tokens: 1024 # Agent interaction patterns interaction_patterns: sequential: description: "Agents work in sequence" pattern: ["planner", "executor", "critic"] parallel: description: "Agents work in parallel" max_concurrent: 3 hierarchical: description: "Coordinator manages other agents" coordinator: "coordinator" workers: ["executor", "memory"] feedback_loop: description: "Iterative improvement with critic" pattern: ["executor", "critic", "executor"] max_iterations: 3