Skip to content

claude-world/director-mode-lite / skills/agent-template

agent-template

Verified

Generate custom agent from template. Use when creating a new subagent from scratch, or scaffolding an agent file with correct frontmatter.

Agent Template Generator

Generate a custom agent file based on requirements.

Usage: /agent-template [agent-name] [purpose]


Templates

PurposeTemplateToolsColorModel
Review/AuditReviewerRead, Grep, Glob, Bashyellowinherit
Generate/CreateGeneratorRead, Write, Grep, Globcyaninherit
Fix/ModifyFixerRead, Write, Edit, Bash, Grepredinherit
Test/ValidateTesterRead, Bash, Grep, Globgreeninherit
DocumentDocumenterRead, Write, Grep, Globcyaninherit
OrchestrateOrchestratorRead, Write, Bash, Grep, Glob, Agentcyanhaiku

Process

  1. Gather Requirements

    • Agent name (lowercase, hyphenated)
    • Purpose
    • Tools needed
    • Model (fable/opus/sonnet/haiku/inherit — inherit is the recommended default)
    • Preloaded skills (if any)
  2. Select Template based on purpose

  3. Generate File at .claude/agents/[name].md

  4. Validate with /agent-check


Frontmatter Reference (official fields)

---
name: agent-name            # Required: lowercase, hyphenated (official: <=256 chars, no ':')
description: >              # Required: 200-1000 chars recommended, include <example> blocks.
  Use this agent PROACTIVELY when [triggering conditions]. Examples:
  <example>
  Context: [situation]
  user: "[request]"
  assistant: "[response]"
  <commentary>[why this agent]</commentary>
  </example>
color: cyan                 # Required by Director Mode convention (CI-enforced); optional per spec
model: inherit              # Required by Director Mode convention (CI-enforced); optional per spec.
                            #   Valid: sonnet, opus, haiku, fable, inherit, sonnet[1m], opus[1m],
                            #   or a full model ID (claude-opus-5-5). inherit recommended.
                            #   /model-only values (default, best, opusplan) are not subagent values.
effort: medium              # Optional: low, medium, high, xhigh, max
tools:                      # Optional: YAML list (omit = all tools)
  - Read
  - Write
  - Edit
  - Bash
  - Grep
  - Glob
  - Agent
  - WebFetch
  - WebSearch
  - NotebookEdit
disallowedTools:            # Optional: explicit tool blocking
  - NotebookEdit
maxTurns: 20                # Optional: max agentic turns; output is marked partial at the cap
skills:                     # Optional: preload full skill content (list, at most 32)
  - linked-skill
memory: project             # Optional: persistent memory scope, one of user, project, local
background: false           # Optional: true keeps the agent in the background even when Claude waits
isolation: worktree         # Optional: run the agent in an isolated git worktree
omitClaudeMd: false         # Optional (2.1.271+): true starts without CLAUDE.md files
autoCompactWindow: 200000   # Optional (2.1.296+): auto-compact earlier, 100000-1000000 tokens
experimental:               # Optional (2.1.248+): prompt cache lifetime, 5m or 1h
  cacheTtl: 5m
---

Ignored for plugin subagents — do not emit these here: hooks, mcpServers, permissionMode, and initialPrompt work in .claude/agents/, but Director Mode also ships its agents as a plugin, where Claude Code ignores them. forkContext is not an official field, and unknown fields are silently ignored.


Reviewer Template Structure

---
name: [name]
description: >
  Use this agent PROACTIVELY when [triggering conditions]. Examples:
  <example>
  Context: [situation]
  user: "[request]"
  assistant: "[response]"
  </example>
color: yellow
tools:
  - Read
  - Grep
  - Glob
  - Bash
model: inherit
skills:
  - linked-skill
# maxTurns: 20
# memory: project
---

# [Name] Agent

## Activation
When to trigger

## Review Checklist
- [ ] Check items

## Output Format
Report structure

Example

/agent-template security-scanner "scan code for vulnerabilities"

Output: Created .claude/agents/security-scanner.md
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 468kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 321kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 321kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 321kUpdated Claude CodeCodexVerifiedInstall