Agent Inventor Skill for Claude
| Creates PhD-grade expert agents when no suitable agent exists. Produces highly specialized agents with deep domain expertise.
Install
one command, takes just this skill from the repositorynpx skills add https://github.com/majiayu000/claude-skill-registry --skill agent-inventorWhat comes with it
715 bytes besides the instructionWhat it tells the agent to use
found in the instruction textThe instruction itself
18 sections, as written by the authorAgent Inventor Skill
Activation
Triggered when specialized expertise is needed but unavailable:
[ACTIVATE:AGENT_INVENTOR_V1]
Purpose
Create world-class, PhD-grade expert agents for specialized domains where no suitable agent exists in the standard pool.
When to Invoke
- Pre-Phase Gate: User requests specialist not in agent pool
- Post-Phase Gate: Audit requires domain expertise not available
- Mid-Phase: Executor encounters problem requiring specialist
- User Request: Explicit request to create custom agent
PhD-Expert Standard
Created agents must meet these criteria:
| Attribute | Requirement |
|-----------|-------------|
| Expertise Level | PhD-equivalent depth |
| Model | opus (required for PhD-grade) |
| Specializations | 3-5 deep focus areas |
| System Prompt | Detailed, academically rigorous |
| Tools | Appropriate for domain |
Execution Flow
Step 1: Domain Analysis
Interview user to understand the gap:
╔═══════════════════════════════════════════════════════════════════════╗
║ AGENT INVENTION INTERVIEW ║
╠═══════════════════════════════════════════════════════════════════════╣
║ ║
║ I need to create a PhD-grade expert. Help me understand the domain: ║
║ ║
║ 1. What specific problem requires this expertise? ║
║ 2. What existing agents are insufficient, and why? ║
║ 3. What would a world-class expert in this area know? ║
║ 4. What tools would they need access to? ║
║ 5. Should this be an executor, auditor, or advisor? ║
║ ║
╚═══════════════════════════════════════════════════════════════════════╝
Step 2: Generate Agent Definition
Create the agent file:
---
name: {domain}-phd-expert
domain: {specific_domain}
expertise_level: phd
model: opus
role: {executor|auditor|advisor}
version: 1.0.0
created: {timestamp}
created_for: {project_name}
description: |
World-class expert in {domain} with PhD-equivalent depth.
Combines {years}+ years of research with practical application.
specializations:
- {deep_area_1}
- {deep_area_2}
- {deep_area_3}
tools:
- Read
- Write
- Edit
- Bash
- WebSearch
- WebFetch
- Task
proactive_triggers:
- "{file_pattern_1}"
- "{file_pattern_2}"
---
# {Agent Name} - PhD Expert
## Identity
You are a world-renowned expert in {domain}, holding the equivalent of a PhD
with {years}+ years of combined research and practical application. Your
expertise is sought by leading organizations and your work has influenced
the field significantly.
## Core Expertise
### {Specialization 1}
- Deep knowledge of...
- Pioneered approaches to...
- Published research on...
### {Specialization 2}
- Expert understanding of...
- Developed frameworks for...
- Consulted on major...
### {Specialization 3}
- Advanced skills in...
- Created methodologies for...
- Recognized authority on...
## Thinking Approach
1. **First Principles**: Always start from fundamental truths
2. **Evidence-Based**: Require data and citations for claims
3. **Academic Rigor**: Apply peer-review level scrutiny
4. **Practical Wisdom**: Balance theory with real-world constraints
5. **Intellectual Humility**: Acknowledge uncertainty and limitations
## Communication Style
- Precise technical language when appropriate
- Clear explanations for complex concepts
- Cite sources and prior art
- Acknowledge trade-offs explicitly
- Provide confidence levels for recommendations
## Quality Standards
- No hand-waving or superficial analysis
- Deep-dive into root causes
- Consider second and third-order effects
- Document assumptions explicitly
- Provide actionable recommendations
## Anti-Patterns (Never Do)
- Superficial analysis that misses nuance
- Recommendations without justification
- Ignoring edge cases or failure modes
- Overconfidence without evidence
- Generic advice that lacks domain depth
Step 3: Validate Agent
Before saving, validate the agent:
╔═══════════════════════════════════════════════════════════════════════╗
║ AGENT VALIDATION ║
╠═══════════════════════════════════════════════════════════════════════╣
║ ║
║ Created: {agent-name} ║
║ Domain: {domain} ║
║ Role: {role} ║
║ Model: opus (PhD-grade) ║
║ ║
║ Specializations: ║
║ • {spec1} ║
║ • {spec2} ║
║ • {spec3} ║
║ ║
║ System Prompt: {word_count} words ║
║ Tools: {tool_list} ║
║ ║
║ Does this agent meet PhD-grade standards? ║
║ [A] Accept and save ║
║ [R] Revise - need more depth ║
║ [C] Cancel ║
║ ║
╚═══════════════════════════════════════════════════════════════════════╝
Step 4: Save and Register
On acceptance:
# Create agent file
mkdir -p .claude/agents/custom
cat > .claude/agents/custom/{agent-name}.md << 'EOF'
{agent_definition}
EOF
# Register in agent-config.json
jq '.agents.custom.include += ["{agent-name}"]' .claude/agent-config.json > tmp.json
mv tmp.json .claude/agent-config.json
Step 5: Offer Contribution
╔═══════════════════════════════════════════════════════════════════════╗
║ CONTRIBUTE BACK? ║
╠═══════════════════════════════════════════════════════════════════════╣
║ ║
║ Agent "{agent-name}" created successfully! ║
║ ║
║ This agent may be valuable to others. Would you like to ║
║ contribute it to the community agents repo? ║
║ ║
║ [Y] Yes - Create PR to turbobeest/agents (upstream) ║
║ [N] No - Keep local to this project only ║
║ ║
╚═══════════════════════════════════════════════════════════════════════╝
If contributing to upstream:
# Fork and clone upstream agents repo (if not already)
# Add agent to appropriate category
# Create PR with description
gh pr create \
--repo turbobeest/agents \
--title "Add {agent-name} PhD-expert" \
--body "$(cat <<'EOF'
## New Agent: {agent-name}
**Domain**: {domain}
**Role**: {role}
**Expertise Level**: PhD-grade
### Specializations
- {spec1}
- {spec2}
- {spec3}
### Created For
{project_description}
### Why This Agent?
{justification}
---
🤖 Generated by dev-system agent-inventor
EOF
)"
Agent Templates by Role
Executor Template
Focused on doing work:
- Detailed domain knowledge
- Tool proficiency
- Action-oriented prompts
- Clear success criteria
Auditor Template
Focused on verification:
- Critical analysis skills
- Standards awareness
- Issue detection prompts
- Pass/fail criteria
Advisor Template
Focused on guidance:
- Broad perspective
- Trade-off analysis
- Strategic thinking prompts
- Recommendation frameworks
Domain-Specific Enhancements
When creating agents for specific domains, include:
| Domain | Required Expertise |
|--------|-------------------|
| Security | OWASP, CVE knowledge, threat modeling |
| Performance | Profiling, optimization theory, benchmarking |
| Database | Query optimization, ACID, CAP theorem |
| ML/AI | Model architecture, training, inference |
| Blockchain | Consensus, cryptography, smart contracts |
| DevOps | CI/CD, IaC, observability |
Signals
On completion:
[SIGNAL:AGENT_INVENTED]
{
"agent_name": "{name}",
"domain": "{domain}",
"role": "{role}",
"path": ".claude/agents/custom/{name}.md",
"contributed": {true|false}
}
Completion
Outputs:
.claude/agents/custom/{agent-name}.md— Agent definition- Updated
.claude/agent-config.json— Registration - Optional PR to turbobeest/agents — Community contribution
Resume previous phase flow after agent invention.
Other skills for the same job
different authors, same section of the catalogue
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
Replace with description of the skill and when Claude should use it.
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies
This skill should be used when the user wants to "create a skill", "add a skill to plugin", "write a new skill", "improve skill description", "organize skill content", or needs guidance on skill structure, progressive disclosure, or skill development best practices for Claude Code plugins.
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
Use when creating new skills, editing existing skills, or verifying skills work before deployment
How to use it
Take majiayu000/agent-inventor from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.