| name | agent-design |
| description | design and review llm agent systems, tool interfaces, and skill boundaries. use when chatgpt needs to decide whether a task should be handled by a prompt, workflow, skill, tool, single-agent loop, or multi-agent system; define tool schemas, instructions, guardrails, evaluation plans, or permission boundaries; or critique an existing agent design for over-orchestration, weak context handling, missing verification, or poor tool design. |
Agent Design Skill
Core Principles
Three axioms everything else derives from:
- Tools are primitive system boundaries. They expose raw capabilities without encoding usage assumptions.
- Skills are usage knowledge. They teach the agent when and how to wield those primitivesโas knowledge, workflows, CLI wrappers, delegation patterns, or tool integration guides.
- The context window is real state. Everything else is a projection.
Quick Decision Rules
Tool or Skill?
- New system boundary (shell, filesystem, web, external API) โ Tool
- Teaching the agent to use an existing capability reliably โ Skill
- See decision-tree.md for full breakdown
Workflow or Agent loop?
- Subtasks are predictable and fixed-order โ Workflow
- Path depends on prior results and can't be hardcoded โ Agent
Single-agent or Multi-agent?
- One context window is sufficient โ Single
- Tasks are genuinely independent, or require specialized context separation โ Multi
- When in doubt, start single. See patterns.md
When to add orchestration?
- Only when simpler structures demonstrably fail. Never for architectural elegance alone.
Red Flags
If you see any of these, the design has likely drifted:
References
