Custom Agent Definitions

Custom Agent Definitions

Expert knowledge for defining and configuring custom agents in Claude Code.

Core Concepts

Custom Agents allow you to define specialized agent types beyond the built-in ones (Explore, Plan, Bash, etc.). Each custom agent can have its own model, tools, and context configuration.

Agent Definition Schema

Custom agents are defined in .claude/agents/ or via plugin agent directories.

Basic Structure

---
name: my-custom-agent
description: What this agent does
model: sonnet
allowed-tools: Bash, Read, Grep, Glob
---

# Agent System Prompt

Instructions and context for the agent...

Context Forking

The context field controls how the agent's context relates to the parent conversation:

Example: Isolated Research Agent

---
name: research-agent
description: Research questions without modifying main context
model: sonnet
context: fork
allowed-tools: WebSearch, WebFetch, Read
---

# Research Agent

You are a research specialist. Search for information and provide findings.
Your research doesn't affect the main conversation context.

When to use context: fork:

• Exploratory research that shouldn't pollute main context
• Parallel investigations with potentially conflicting approaches
• Isolated experiments or testing
• Background tasks that run independently

Agent Field for Delegation

The agent field specifies which agent type to use when delegating via the Task tool:

---
name: code-review-workflow
description: Comprehensive code review
agent: security-auditor
allowed-tools: Read, Grep, Glob, TodoWrite
---

This allows commands and skills to specify a preferred agent type for delegation.

Disallowed Tools (Restrictions)

The disallowedTools field explicitly prevents an agent from using certain tools:

---
name: read-only-explorer
description: Explore codebase without modifications
model: haiku
allowed-tools: Bash, Read, Grep, Glob
disallowedTools: Write, Edit, NotebookEdit
---

# Read-Only Explorer

Explore and analyze code. Do not make any modifications.

Disallowed Tools vs Allowed Tools:

When to use disallowedTools:

• Creating read-only agents that can explore but not modify
• Restricting dangerous capabilities (Bash execution)
• Sandboxing agents for specific tasks
• Security-sensitive contexts

Complete Example

---
name: security-auditor
description: Security-focused code review agent
model: sonnet
context: fork
allowed-tools: Read, Grep, Glob, WebSearch, TodoWrite
disallowedTools: Bash, Write, Edit
created: 2026-01-20
modified: 2026-01-20
reviewed: 2026-01-20
---

# Security Auditor Agent

You are a security auditor. Analyze code for vulnerabilities.

## Capabilities
- Read and analyze source code
- Search for security patterns
- Research known vulnerabilities
- Track findings in todo list

## Restrictions
- Cannot execute code (no Bash)
- Cannot modify files (no Write/Edit)
- Work in isolated context

## Focus Areas
1. SQL injection vulnerabilities
2. XSS vulnerabilities
3. Authentication/authorization flaws
4. Secrets/credentials in code
5. Insecure dependencies

Defining Agents in Plugins

Plugins can define custom agents in their agents/ directory:

my-plugin/
├── .claude-plugin/
│   └── plugin.json
├── agents/
│   ├── security-auditor.md
│   ├── performance-analyzer.md
│   └── accessibility-checker.md
└── skills/
    └── ...

Each agent file follows the same YAML frontmatter + markdown body structure.

Using Custom Agents

Via Task Tool

Task tool with subagent_type="security-auditor" for security analysis.

Via Delegation

/delegate Audit auth module for security issues

The delegation system matches tasks to appropriate custom agents.

Best Practices

1. Principle of Least Privilege

Only grant tools the agent actually needs:

# Good: Minimal tools for the task
allowed-tools: Read, Grep, Glob

# Avoid: Overly permissive
allowed-tools: Bash, Read, Write, Edit, Grep, Glob, WebSearch, WebFetch

2. Use Context Forking for Isolation

# Good: Isolated exploratory work
context: fork

3. Combine Allowed and Disallowed

# Explicit whitelist with safety blacklist
allowed-tools: Bash, Read, Grep
disallowedTools: Write, Edit

4. Clear Agent Descriptions

description: |
  Security auditor for identifying vulnerabilities in authentication
  and authorization code. Reports findings without modifying code.

5. Model Selection

Agent Configuration Fields Reference

Common Patterns

Read-Only Research Agent

context: fork
allowed-tools: Read, Grep, Glob, WebSearch, WebFetch
disallowedTools: Bash, Write, Edit

Safe Code Executor

allowed-tools: Bash, Read
disallowedTools: Write, Edit

Documentation Writer

allowed-tools: Read, Write, Edit, Grep, Glob
disallowedTools: Bash

Full-Power Developer

allowed-tools: Bash, Read, Write, Edit, Grep, Glob, TodoWrite

Agentic Optimizations

Quick Reference

Context Modes

Tool Restriction Patterns