Agent Catalog

Agent System

oh-my-claudecode has 19 specialized agents. Each agent has a defined role and is invoked using the format oh-my-claudecode:agent-name.

The orchestrator automatically delegates to the appropriate agent based on the nature of the task, and users can also specify an agent directly.

4-Lane Structure

Agents are divided into 4 lanes based on their roles.

Build & Analysis

Covers the full development flow from codebase exploration to implementation and verification.

Agent	Default Model	Role
`explore`	haiku	Codebase exploration, file/symbol search
`analyst`	opus	Requirements analysis, acceptance criteria definition
`planner`	opus	Task sequencing, execution plan creation
`architect`	opus	System design, boundaries, interface definition
`debugger`	sonnet	Root cause analysis, build error resolution
`executor`	sonnet	Code implementation, refactoring
`verifier`	sonnet	Completion evidence collection, claim validation
`tracer`	sonnet	Evidence-based causal tracing, competing hypotheses

Review

Reviews code quality and security.

Agent	Default Model	Role
`security-reviewer`	sonnet	OWASP Top 10, authentication/authorization vulnerability review
`code-reviewer`	opus	Logic defects, API contracts, backward compatibility review

Domain

Handles domain-specific specialized tasks.

Agent	Default Model	Role
`test-engineer`	sonnet	Test strategy, coverage, TDD
`designer`	sonnet	UI/UX architecture, interaction design
`writer`	haiku	Technical docs, README, API documentation
`qa-tester`	sonnet	CLI/service runtime validation via tmux
`scientist`	sonnet	Data/statistical analysis, research execution
`git-master`	sonnet	Atomic commits, rebase, history management
`document-specialist`	sonnet	External documentation/reference lookup
`code-simplifier`	opus	Code clarity and consistency improvement

Coordination

Applies critical review to plans and designs.

Agent	Default Model	Role
`critic`	opus	Critical review of plans/designs, multi-perspective analysis

Model Tiers

Each agent has a default model tier assigned, but it can be changed at call time using the model parameter.

Tier	Model	Characteristics	Representative Agents
LOW	haiku	Fast and cheap. Suitable for simple lookups	explore, writer
MEDIUM	sonnet	Balanced performance. Suitable for implementation/debugging	executor, debugger, test-engineer
HIGH	opus	Top-level reasoning. Suitable for design/analysis	architect, planner, critic, code-reviewer

Some agents support tier variants. For example, explore defaults to haiku, but explore-high uses the opus tier.

Agent Selection Guide

Task Type	Recommended Agent	Model
Quick code lookup	`explore`	haiku
Find files/patterns	`explore`	haiku
Feature implementation	`executor`	sonnet
Complex refactoring	`executor` (model=opus)	opus
Simple bug fix	`debugger`	sonnet
Complex debugging	`architect`	opus
UI components	`designer`	sonnet
Write documentation	`writer`	haiku
SDK/API doc lookup	`document-specialist`	sonnet
Test strategy	`test-engineer`	sonnet
Security review	`security-reviewer`	sonnet
Code review	`code-reviewer`	opus
Data analysis	`scientist`	sonnet
Strategic planning	`planner`	opus
Plan review	`critic`	opus

Agent Workflow

In a typical task, agents collaborate in the following order.

explore (discovery)
  → analyst (requirements analysis)
    → planner (plan creation)
      → critic (plan review)
        → executor (implementation)
          → verifier (verification)

Detailed descriptions of each agent are available on the sub-pages within each lane.

Deprecated Agent Aliases

The following aliases remain for backward compatibility.

Alias	Actual Agent
`api-reviewer`	code-reviewer
`performance-reviewer`	code-reviewer
`quality-reviewer`	code-reviewer
`quality-strategist`	code-reviewer
`dependency-expert`	document-specialist
`researcher`	document-specialist
`tdd-guide`	test-engineer
`deep-executor`	executor
`build-fixer`	debugger
`harsh-critic`	critic

On this page