| name | agent-designer |
| description | Design multi-step AI agent architectures with tools, memory, and orchestration.
TRIGGERS - Use when user wants to design AI agents, multi-agent systems, or autonomous AI workflows.
|
AI Agent Designer
Overview
Designs AI agent architectures — single agents or multi-agent systems — with clear roles, tools, memory, and orchestration patterns.
Workflow
Step 1: Define the Agent
- Objective: What should the agent accomplish?
- Scope: What's in and out of bounds?
- Tools needed: What APIs/services should it access?
- Autonomy level: Fully autonomous, human-in-the-loop, or supervised?
- Platform: LangChain, CrewAI, AutoGen, custom, or conceptual?
Step 2: Choose the Architecture
| Pattern | When to Use |
|---|
| Single agent + tools | One task, multiple steps |
| Sequential chain | Linear workflow, each step feeds the next |
| Router agent | Different tasks need different specialists |
| Parallel agents | Independent tasks that can run simultaneously |
| Hierarchical | Manager delegates to worker agents |
| Collaborative | Agents discuss and refine outputs |
Step 3: Design the System
For each agent:
## Agent: [Name]
### Role
[One sentence — what this agent does]
### System Prompt
[Full system prompt]
### Tools Available
| Tool | Purpose | Input | Output |
|------|---------|-------|--------|
| [tool] | [why] | [format] | [format] |
### Memory
- **Short-term**: [conversation context]
- **Long-term**: [persistent storage approach]
- **Shared**: [what other agents can access]
### Decision Logic
- IF [condition] → [action]
- IF [condition] → [escalate to human]
- IF [error] → [fallback]
### Output
[What this agent produces and where it goes]
Output Format
# AI Agent System: [Name]
## Architecture Overview
**Pattern**: [architecture type]
**Agents**: [count]
**Orchestration**: [how they coordinate]
## System Diagram
[Text-based architecture diagram]
## Agent Specifications
### Agent 1: [Name]
[Full specification per template above]
### Agent 2: [Name]
[Full specification]
## Orchestration Logic
[How agents communicate and coordinate]
## Error Handling
[What happens when things go wrong]
## Human Touchpoints
[Where and when humans intervene]
## Implementation Guide
- **Platform**: [recommendation]
- **Models**: [which LLMs for which agents]
- **Estimated cost**: [per-run cost estimate]
- **Setup steps**: [how to build it]
Quality Checklist
