How do I install the Knowledge Agent Design skill?

Run npx skillmds@latest add deankelly751/knowledge-agent-design in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Knowledge Agent Design safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Knowledge Agent Design?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Knowledge Agent Design free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Knowledge Agent Design?

DeanKelly751 (@deankelly751) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Knowledge Agent Design by deankelly751

Name: Knowledge Agent Design
Author: DeanKelly751

Knowledge Agent Design

AI agent architecture patterns covering tool use, planning, memory, and multi-agent coordination. Decision framework for agent design choices. Use when designing AI agent systems, choosing agent architectures, or reviewing agent implementations.

DeanKelly751 Updated Sep 21, 2026

File contents

Agent Design

Purpose

Decision frameworks for designing effective AI agent systems.

Agent Architecture Patterns

Pattern	Complexity	Best for
Single-turn tool use	Low	Straightforward tasks with clear tool mapping
ReAct (Reason + Act)	Medium	Multi-step tasks requiring planning
Plan-then-execute	Medium	Tasks where the full plan can be determined upfront
Reflexion/self-critique	High	Tasks requiring quality iteration
Multi-agent collaboration	High	Complex tasks benefiting from specialization

Tool Design Principles

Atomic tools: Each tool does one thing well
Clear interfaces: Typed inputs/outputs, explicit constraints
Idempotent where possible: Safe to retry on failure
Observable: Return enough context for the agent to verify success
Fail informatively: Error messages that guide correction

Memory Patterns

Type	Duration	Use for
Context window	Single conversation	Working state
Conversation summary	Session	Long interactions
Retrieval (RAG)	Persistent	Large knowledge bases
Structured memory	Persistent	Facts, preferences, state

Agent Failure Modes

Failure	Cause	Mitigation
Infinite loops	No termination condition	Max iterations, loop detection
Tool misuse	Ambiguous tool descriptions	Better descriptions, examples
Hallucinated actions	No validation of outputs	Output validation, guardrails
Context overflow	Too much history	Summarization, relevance filtering
Cascading errors	Unchecked failure propagation	Error boundaries, rollback

Design Checklist

Clear termination conditions
Maximum iteration limits
Error recovery strategy
Output validation/guardrails
Fallback to human when uncertain
Observability (log decisions, not just actions)

Related Skills

ai-workflows/knowledge-prompt-engineering — prompting agents effectively
ai-workflows/knowledge-guardrails — safety and validation
ai-workflows/execution-rag-pipeline — retrieval for agent knowledge

---name: knowledge-agent-designdescription: AI agent architecture patterns covering tool use, planning, memory, and multi-agent coordination. Decision framework for agent design choices. Use when designing AI agent systems, choosing agent architectures, or reviewing agent implementations.---# Agent Design## PurposeDecision frameworks for designing effective AI agent systems.## Agent Architecture Patterns| Pattern | Complexity | Best for ||---------|-----------|----------|| Single-turn tool use | Low | Straightforward tasks with clear tool mapping || ReAct (Reason + Act) | Medium | Multi-step tasks requiring planning || Plan-then-execute | Medium | Tasks where the full plan can be determined upfront || Reflexion/self-critique | High | Tasks requiring quality iteration || Multi-agent collaboration | High | Complex tasks benefiting from specialization |## Tool Design Principles- **Atomic tools**: Each tool does one thing well- **Clear interfaces**: Typed inputs/outputs, explicit constraints- **Idempotent where possible**: Safe to retry on failure- **Observable**: Return enough context for the agent to verify success- **Fail informatively**: Error messages that guide correction## Memory Patterns| Type | Duration | Use for ||------|----------|---------|| Context window | Single conversation | Working state || Conversation summary | Session | Long interactions || Retrieval (RAG) | Persistent | Large knowledge bases || Structured memory | Persistent | Facts, preferences, state |## Agent Failure Modes| Failure | Cause | Mitigation ||---------|-------|------------|| Infinite loops | No termination condition | Max iterations, loop detection || Tool misuse | Ambiguous tool descriptions | Better descriptions, examples || Hallucinated actions | No validation of outputs | Output validation, guardrails || Context overflow | Too much history | Summarization, relevance filtering || Cascading errors | Unchecked failure propagation | Error boundaries, rollback |## Design Checklist- [ ] Clear termination conditions- [ ] Maximum iteration limits- [ ] Error recovery strategy- [ ] Output validation/guardrails- [ ] Fallback to human when uncertain- [ ] Observability (log decisions, not just actions)## Related Skills- `ai-workflows/knowledge-prompt-engineering` — prompting agents effectively- `ai-workflows/knowledge-guardrails` — safety and validation- `ai-workflows/execution-rag-pipeline` — retrieval for agent knowledge

Frequently asked questions