skills.rest
okwinds
okwindsCOMMUNITY
Explore all skills by this author →
GitHub

ai-agent-prd

Generate AI Agent PRDs with identity, capability, safety, and evaluation frameworks.

50|3|Updated Jan 30, 2026
Search Tags:#architecture#llm#ai-agent#prd#skeleton#prds
One-click install
npx skills add https://github.com/okwinds/miscellany --skill ai-agent-prd
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ai-agent-prd
Source: https://github.com/okwinds/miscellany/tree/main/agent/skills/ai-agent-prd
Command: npx skills add https://github.com/okwinds/miscellany --skill ai-agent-prd

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This skill enables teams to generate comprehensive AI Agent PRDs that define identity, capability architecture, safety guardrails, and evaluation frameworks for AI agents. It applies to conversational, autonomous, and multi-agent systems, guiding architecture decisions, tool usage, RAG integration, and verification. The skill relies on the repository's SKILL.md and references/templates to guide implementation.

Core Features & Use Cases

  • PRD skeleton generation for AI agent products (identity, capabilities, tools, memory, knowledge, workflows, and safety)
  • Domain-specific guidance and templates drawn from references to tailor PRDs for conversational, autonomous, and multi-agent systems
  • Quality gates including safety guardrails, evaluation rubrics, and golden-conversation templates to drive validation

Quick Start

  1. Run the skeleton generator to produce the standard PRD directory structure bash scripts/generate_agent_prd_skeleton.sh ./docs/agent-prd "Customer Support Agent"
  2. Open AGENT_PRD.md to begin with the executive summary
  3. Populate IDENTIY.md, USE_CASES.md, SKILLS.md, TOOLS.md, MEMORY.md, KNOWLEDGE.md, WORKFLOWS.md, BEHAVIOR.md, SAFETY.md, and EVALUATION.md
  4. Reference references/ (templates and checklists) to ensure completeness
  5. Validate with CHECKLIST.md before handoff

Frequently Asked Questions about ai-agent-prd

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate an AI Agent PRD for conversational and autonomous systems?▼

To generate an AI Agent PRD, run the skeleton generator script to create a standard directory structure covering identity, capabilities, tools, memory, knowledge, workflows, and safety for conversational and autonomous systems.

What should be included in an AI agent product requirements document?▼

An AI agent product requirements document should include identity, capability architecture, safety guardrails, evaluation frameworks, tool usage, RAG integration, and verification details to guide architecture decisions.

Can I use this PRD skeleton for multi-agent systems?▼

Yes, this PRD skeleton supports multi-agent systems by providing domain-specific guidance and templates that tailor architecture decisions, tool usage, and verification for multi-agent workflows.

What's the best way to define safety guardrails and evaluation rubrics for LLM agents?▼

The best way to define safety guardrails and evaluation rubrics for LLM agents is to populate the SAFETY.md and EVALUATION.md files using the provided quality gates and golden-conversation templates.

How do I validate an AI agent PRD before handoff?▼

To validate an AI agent PRD before handoff, reference the repository checklists to ensure completeness and run validation against the CHECKLIST.md file provided in the generated skeleton structure.

Do I need to manually configure RAG integration and memory in my AI agent architecture?▼

You need to manually populate the MEMORY.md and KNOWLEDGE.md files to define RAG integration and memory architecture, using the repository references and templates to guide your implementation decisions.

Related Skills

Explore More Product & Management Skills→
Imad-OuteImad-Oute
2

mh-lw-wiki-init

Creates directory structure, schema, and content catalog for a new LLM Wiki domain.

Community
Advanced
RemseyMailjardRemseyMailjard

prd-to-plan

Convert a PRD into a phased implementation plan with tracer bullets.

Community
Advanced
abdullah1854abdullah1854
1

agent-evaluation

Develop evaluation frameworks for non-deterministic agent systems.

Community
Advanced
zerAdazerAda

bmad-validate-prd

Validate Product Requirements Documents for comprehensiveness, organization, and cohesion.

Community
Intermediate