AI_template/agents

Agent Profiles

This directory contains specialized AI agent profiles. Each profile defines a role, principles, constraints, and workflow for a specific domain.

Available Agents

Agent	File	Model	Use When
Planner	planner.md	opus	Breaking down tasks, planning implementations, risk assessment
Security Auditor	security-auditor.md	opus	Security review, vulnerability assessment, auth flows
Code Reviewer	code-reviewer.md	opus	Code quality, PR review, deep analysis
Frontend Architect	frontend-architect.md	sonnet	UI components, performance, accessibility, React/Next.js
Backend Architect	backend-architect.md	sonnet	System design, databases, APIs, scalability
Test Engineer	test-engineer.md	sonnet	Test strategy, automation, CI/CD, coverage
Documentation Expert	documentation-expert.md	sonnet	Technical writing, user/admin guides, docs maintenance
Build Error Resolver	build-error-resolver.md	sonnet	Fix build/type/lint errors with minimal changes
Loop Operator	loop-operator.md	sonnet	Monitor autonomous loops, detect stalls, escalate
Prompt Engineer *	prompt-engineer.md	sonnet	LLM prompts, agent instructions (optional, for AI projects)

Model Selection

• opus — Deep reasoning: planning, security audit, code review. Slower but more thorough.
• sonnet — Implementation: architecture execution, testing, writing, fixing. Faster turnaround.

Principle: planner does the deep thinking, implementation agents execute the plan.

Tool Restrictions

Each agent declares a tools array in its frontmatter, following the principle of least privilege:

• Read-only agents (planner): Read, Glob, Grep — plans, doesn't implement
• Implementation agents (architects, test-engineer, build-error-resolver): Read, Glob, Grep, Edit, Write, Bash
• Review agents (code-reviewer, security-auditor): Read, Glob, Grep, Bash (for git/scan commands)

Agent Selection

See RULES.md sections 4-5 for the selection protocol and multi-agent coordination.

Using context7 (Shared Guidelines)

All agents use context7 to access up-to-date documentation. Training data may be outdated — always verify through context7 before making recommendations.

When to Use

Always query context7 before:

• Recommending specific library/framework versions
• Suggesting API patterns or method signatures
• Advising on security configurations or CVEs
• Checking for deprecated features or breaking changes
• Verifying browser support or compatibility matrices

How to Use

1. Resolve library ID: Use resolve-library-id to find the correct context7 library identifier
2. Query documentation: Use query-docs with the resolved ID and a specific topic

Example

User asks about React Server Components

1. resolve-library-id: "react" → get library ID
2. query-docs: topic="Server Components patterns"
3. Base recommendations on returned documentation, not training data

What to Verify

Category	Verify
Versions	LTS versions, deprecation timelines, migration guides
APIs	Current method signatures, new features, removed APIs
Security	CVE advisories, security best practices, auth patterns
Performance	Current optimization techniques, benchmarks, configuration
Compatibility	Version compatibility matrices, breaking changes

Critical Rule

When context7 documentation contradicts training knowledge, trust context7. Technologies evolve rapidly — training data may reference deprecated patterns or outdated versions.

Adding a New Agent

1. Create a new .md file in this directory
2. Use consistent frontmatter: name, model, tools, and description
   • model: opus for reasoning-heavy tasks, sonnet for implementation
   • tools: minimal set needed (principle of least privilege)
3. Follow the structure: Role → Core Principles → Constraints → Workflow → Responsibilities → Output Format → Pre-Response Checklist
4. Reference this README for context7 usage instead of duplicating the section
5. Update DOCS.md and README.md to list the new agent