# Create Agent > Generate specialized agents with smart prefix-based naming following established patterns Use when this capability is needed. - Skill: `tomevault-io/create-agent-6` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/create-agent-6` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/create-agent-6/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/create-agent-6 --- # ⚡⚡⚡ CRITICAL REQUIREMENT: CREATE PERFECTLY FORMATTED SPECIALIZED AGENT! ⚡⚡⚡ **THIS IS NOT A TEMPLATE COPYING TASK - THIS IS A CUSTOM AGENT GENERATION TASK!** When you run `/create-agent`, you are REQUIRED to: 1. **ANALYZE** the user's agent requirements from $ARGUMENTS thoroughly 2. **CLASSIFY** agent category and apply smart prefix-based naming automatically 3. **STUDY** existing agent patterns to ensure perfect consistency 4. **GENERATE** a complete specialized agent following exact established style 5. **USE MULTIPLE AGENTS** for comprehensive agent creation: - Spawn one agent to analyze existing agent patterns and templates - Spawn another to classify category and determine optimal prefix - Spawn more agents for template selection and specialization - Say: "I'll spawn multiple agents to create this specialized agent with perfect consistency" 6. **DO NOT STOP** until: - ✅ Agent follows exact format and style of existing agents - ✅ Category classification and prefix assignment are correct - ✅ ALL required sections are present and properly formatted - ✅ Agent is contextually appropriate for the specified purpose - ✅ YAML front matter and markdown structure are perfect - ✅ Task tool integration patterns are properly implemented **FORBIDDEN BEHAVIORS:** - ❌ "Close enough to existing style" → NO! Follow the pattern EXACTLY! - ❌ "Skip some sections to save time" → NO! Include ALL required sections! - ❌ "Generic template will work" → NO! Create contextually specific content! - ❌ "Wrong category prefix assignment" → NO! Analyze description thoroughly! - ❌ Inconsistent formatting → NO! Match existing agents precisely! **MANDATORY WORKFLOW:** ``` 1. Requirements analysis → Parse user's agent description 2. IMMEDIATELY spawn agents for category classification and generation 3. CATEGORY DETECTION → Apply smart prefix assignment 4. Pattern study → Analyze existing agent structure exactly 5. Template selection → Choose appropriate base template 6. Custom generation → Create agent specific to user requirements 7. VERIFY format → Ensure perfect consistency with existing style ``` **YOU ARE NOT DONE UNTIL:** - Agent follows exact format of existing agents - Category classification and prefix assignment are correct - All required sections are present and properly formatted - Content is contextually appropriate for the specified purpose - Agent is saved to **templates/agents/** directory - Task tool integration patterns are properly implemented - Agent can be spawned immediately via Task tool --- 🛑 **MANDATORY AGENT GENERATION PROTOCOL** 🛑 1. Re-read ~/.claude/CLAUDE.md RIGHT NOW 2. Check current TODO.md status 3. **SCAN templates/agents/ directory** for existing agent patterns 4. **APPLY category classification logic** to determine prefix and template 5. Analyze existing agents for pattern consistency Execute comprehensive custom agent generation with ZERO tolerance for format deviations. **FORBIDDEN SHORTCUT PATTERNS:** - "The basic structure is similar" → NO, follow the EXACT pattern - "Some sections don't apply" → NO, adapt sections to fit the purpose - "Users will understand" → NO, maintain strict formatting consistency - "Close enough formatting" → NO, precision is required - "Generic agents work fine" → NO, create specific, specialized agents You are creating a specialized agent for: $ARGUMENTS Let me ultrathink about generating a perfectly formatted specialized agent that follows all established patterns. 🚨 **REMEMBER: Agent consistency is crucial for ecosystem integration!** 🚨 **Comprehensive Specialized Agent Generation Protocol:** **Step 0: Agent Requirements Analysis** Parse the agent description from $ARGUMENTS to identify: - Primary agent purpose and specialized capability - Target domain or process to be automated - Required coordination patterns and integration points - Task tool spawning requirements - Category classification for smart prefix assignment **Step 1: Smart Category Classification and Prefix Assignment** **AGENT CATEGORY CLASSIFICATION:** Analyze description for category keywords and apply smart prefix assignment: **Testing Agents** (prefix: `test-`, `unit-`, `integration-`): - Keywords: "test", "testing", "unit", "integration", "coverage", "spec", "validation" - Template base: `test-orchestrator.md` or `unit-test-master.md` **Milestone Agents** (prefix: `milestone-`): - Keywords: "milestone", "planning", "execution", "coordination", "project", "roadmap" - Template base: `milestone-coordinator.md` **Code Quality Agents** (prefix: `code-`, `quality-`): - Keywords: "quality", "lint", "format", "analyze", "review", "standards", "refactor" - Template base: `code-quality-enforcer.md` **Infrastructure Agents** (prefix: `git-`, `deploy-`, `build-`, `ci-`): - Keywords: "git", "deploy", "build", "ci", "infrastructure", "automation", "pipeline" - Template base: `git-operator.md` **API/Service Agents** (prefix: `api-`, `service-`): - Keywords: "api", "endpoint", "service", "integration", "rest", "graphql", "backend" - Template base: `api-integration-tester.md` **Generic Agents** (prefix: `utility-` or no prefix): - No clear category match or explicitly generic purpose - Template base: `orchestrator.md` **Step 2: Pattern Analysis and Template Extraction** Study existing agents to extract exact patterns: - YAML front matter format and required fields - Emoji usage patterns and intensity indicators - Section headers and formatting conventions - Mission statement structure and mandatory elements - Task tool spawning patterns and coordination logic - Quality gates and verification criteria **Step 3: Multi-Agent Specialized Generation Strategy** Deploy specialized generation agents for parallel development: ``` "I need to create a [CATEGORY] agent with smart prefix naming. I'll spawn generation agents: - Agent 1: Analyze existing agent patterns and extract template structure - Agent 2: Apply category classification and suggest optimal prefix - Agent 3: Select appropriate template base and customize for purpose - Agent 4: Generate agent-specific YAML front matter and core mission - Agent 5: Create Task tool spawning patterns and coordination logic - Agent 6: Validate against existing agents and ensure naming consistency Let me generate this specialized agent with perfect pattern compliance..." ``` **Agent Structure Requirements:** **AGENT YAML Front Matter (MANDATORY):** ```yaml --- name: [category-prefix-name] description: [Specific agent purpose with use cases and examples] model: sonnet --- ``` **Agent Structure Pattern:** - **Mission Statement** (🎯 CORE MISSION:) - **Parallel Agent Deployment** (🚀 TRUE PARALLELISM VIA TASK TOOL) - **Core Capabilities** (📊, 🔧, 🧠 sections) - **Coordination Patterns** (📈, 🔄 sections) - **Quality Gates** (✅, 🚨 sections) - **Task Tool Integration** with standardized spawning patterns **Core Requirements Section:** - Numbered list (1-5 items) of specific actions required - "USE MULTIPLE AGENTS" subsection with examples - "DO NOT STOP" until criteria with ✅ checkmarks **Forbidden Behaviors Section:** - ❌ pattern with specific examples relevant to agent purpose - Common mistakes and anti-patterns to avoid **Mandatory Workflow:** - Code block with numbered steps - Clear progression from start to completion **Completion Criteria:** - "YOU ARE NOT DONE UNTIL:" section - Specific, measurable completion requirements **Main Protocol Section:** - Detailed step-by-step execution protocol - Step 0, 1, 2, etc. with specific requirements - Checklists with [ ] checkbox format - Quality requirements and standards **Task Tool Spawning Strategy:** - Specific examples of agent deployment - Clear division of responsibilities - Parallel execution patterns **Anti-patterns Section:** - ❌ forbidden behaviors specific to agent purpose - Clear guidance on what NOT to do **Final Sections:** - Verification criteria for completion - Final commitment with "I will" and "I will NOT" lists - Remember statement - Execution statement **Agent Types and Adaptation Patterns:** **Testing Agents** (🧪🧪🧪): - Focus on comprehensive test orchestration - Multiple testing agents for different aspects - Test coverage and validation protocols - Framework-specific testing strategies **Milestone Agents** (🎯🎯🎯): - Emphasis on project coordination and execution - Parallel milestone agents for planning/execution - Progress tracking and delivery gates - Strategic planning requirements **Code Quality Agents** (🔍🔍🔍): - Zero-tolerance for quality issues - Comprehensive analysis and enforcement - Standards compliance requirements - Code improvement protocols **Step 4: Content Generation and Specialization** Generate agent content specifically tailored to user requirements: - Adapt emoji intensity to agent criticality - Create contextually relevant forbidden behaviors - Generate appropriate workflow steps - Develop domain-specific checklists - Create relevant Task tool spawning examples **Content Quality Standards:** - [ ] All sections present and properly formatted - [ ] Content is specific to agent purpose (not generic) - [ ] Examples are contextually relevant and actionable - [ ] Formatting matches existing agents exactly - [ ] YAML front matter is valid and complete - [ ] Markdown structure follows established patterns **Step 5: Format Validation and Consistency Check** Verify the generated agent against existing patterns: - [ ] YAML front matter format is identical - [ ] Section headers follow exact capitalization and formatting - [ ] Emoji patterns match intensity and purpose - [ ] Workflow structure follows established format - [ ] Task tool spawning examples follow established patterns - [ ] Checklist formatting is consistent - [ ] Final sections follow exact order and format **Step 6: Intelligent Agent Placement Strategy** **AGENT PLACEMENT:** - **Target directory**: `templates/agents/` (NOT templates/skills/) - **Naming convention**: Apply detected category prefix + descriptive name - **Validation**: Check against existing agent names for uniqueness - **Examples**: - Testing agent: `test-performance.md`, `unit-coverage.md` - Milestone agent: `milestone-tracker.md` - Code quality: `code-formatter.md`, `quality-auditor.md` - Infrastructure: `git-automation.md`, `deploy-coordinator.md` **File Generation and Placement:** - Generate appropriate filename based on category classification - Save to **templates/agents/** directory - Apply smart prefix naming before file creation - Ensure file permissions are correct - Validate file was created successfully in correct location **Agent Quality Checklist:** - [ ] Follows exact YAML front matter format (name, description, model) - [ ] Category prefix applied correctly based on classification - [ ] Name is unique and doesn't conflict with existing agents - [ ] Mission statement follows 🎯 CORE MISSION pattern - [ ] Task tool integration patterns included - [ ] Parallel agent deployment section present - [ ] Core capabilities properly structured (📊, 🔧, 🧠) - [ ] Coordination patterns included (📈, 🔄) - [ ] Quality gates defined (✅, 🚨) - [ ] File saved to templates/agents/ directory **Agent Creation Anti-patterns (FORBIDDEN):** - ❌ "Wrong category prefix assignment" → NO, analyze description thoroughly - ❌ "Skip Task tool integration patterns" → NO, all agents must spawn properly - ❌ "Generic agent without specialization" → NO, create domain-specific agents - ❌ "Duplicate existing agent functionality" → NO, validate uniqueness first - ❌ "No coordination patterns" → NO, agents must integrate with ecosystem - ❌ "Copy existing agent exactly" → NO, customize for specific purpose **Template Adaptation Guidelines:** - Maintain exact structure while adapting content - Use domain-specific language and examples - Ensure forbidden behaviors are contextually relevant - Create appropriate success criteria for the purpose - Generate realistic Task tool spawning scenarios **Final Agent Generation Verification:** **Agent Completion Criteria:** ✓ Category classification applied correctly with appropriate prefix ✓ Agent follows exact structure of existing agent templates ✓ YAML front matter includes name, description, model fields ✓ Mission statement and core capabilities clearly defined ✓ Task tool integration patterns properly implemented ✓ Coordination with existing agent ecosystem validated ✓ File is properly saved to templates/agents/ ✓ Agent can be spawned immediately via Task tool **Final Agent Generation Commitment:** I will execute EVERY generation step and CREATE PERFECT SPECIALIZED AGENT. I will: - ✅ Apply smart category classification with appropriate prefixes - ✅ Analyze agent requirements thoroughly - ✅ SPAWN MULTIPLE AGENTS for pattern analysis and generation - ✅ Study existing agents to extract exact patterns - ✅ Generate contextually specific and appropriately formatted agent - ✅ Verify perfect consistency with established style and naming conventions - ✅ Save completed agent to templates/agents/ directory I will NOT: - ❌ Create generic or template-based agents - ❌ Skip category classification or prefix assignment steps - ❌ Apply wrong prefixes or ignore naming conventions - ❌ Skip any required sections or formatting - ❌ Accept "close enough" pattern matching - ❌ Generate agents that don't follow exact style - ❌ Leave any formatting inconsistencies - ❌ Stop working while any format deviations remain **REMEMBER: This is a PERFECT SPECIALIZED AGENT GENERATION task, not template copying!** The agent is complete ONLY when it follows exact patterns, uses correct naming conventions with smart prefixes, and serves the specific purpose perfectly. **Executing comprehensive specialized agent generation NOW...** --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/nashgao) — claim your Tome and manage your conversions.