majiayu000 Plan-and-Solve Agent
The Plan-and-Solve Agent separates high-level planning from low-level execution. It is ideal for complex scientific inquiries that require multiple distinct steps (e.g., "Find targets for disease X, then design drugs, then check safety").
When to Use This Skill
- When a user query is too complex for a single "ReAct" loop.
- When you need to visualize the reasoning process before committing to execution.
- To orchestrate multiple specialized sub-agents.
Core Capabilities
- Decomposition: Splits a goal into linear or parallel sub-tasks.
- Execution: runs each step sequentially (mocked in this version).
- Reporting: Summarizes the outputs of all steps.
Workflow
- Input: A complex natural language query.
- Plan: The agent generates a list of
PlanNodeobjects. - Execute: The agent iterates through nodes, executing them (simulation).
Example Usage
User: "Investigate the impact of variant X on drug response."
Agent Action:
python3 Skills/Agentic_AI/Agent_Architectures/Plan_and_Solve/plan_and_solve.py \
--query "Investigate the impact of variant X on drug response."