Skip to content
Get packs

Agent Development Guide

Lets the agent create well-structured Claude Code subagent files with triggers, prompts and tool limits

Source Download skill
Repository
anthropics/claude-plugins-official
Stars
45,770
Licence
Apache-2.0
Files
7 files

What it does

When you ask for help creating a Claude Code subagent (a specialised helper that runs a multi-step task on its own), this skill gives the agent the file format, naming rules, field options and writing guidance to follow. It explains how to write the description that tells Claude when to use the agent, how to write the system prompt, and how to limit which tools the agent can use. It also includes a script that checks an agent file for correct structure.

Great for

  • Plugin developers who are building custom Claude Code subagents
  • Teams standardising how their internal agents are written and validated

Skip it if

  • Non-developers who only want to use an existing agent, since this skill is about authoring them
  • Simple one-off prompts that do not need a reusable autonomous helper

Try asking

  • “Create an agent that reviews pull requests for security issues”
  • “Write a subagent for generating API documentation with limited tools”
  • “Check my agent markdown file for missing frontmatter fields”
  • “Help me write a good description so Claude triggers my test-generator agent at the right time”
Step-by-step guide: how to use Agent Development GuideInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 7 original files, the Apache-2.0 licence and an attribution note.

    agent-development.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip agent-development.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip agent-development.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Run scripts/validate-agent.sh on any agent file you write before publishing it in a plugin
  • Keep tool access to the minimum needed, for example read-only tools for review agents
  • Review generated system prompts yourself, since the skill suggests a generation pattern that still needs human checking

What's inside

  • examples/agent-creation-prompt.md
  • examples/complete-agent-examples.md
  • references/agent-creation-system-prompt.md
  • references/system-prompt-design.md
  • references/triggering-examples.md
  • scripts/validate-agent.sh
  • SKILL.md

Redistributed unmodified from anthropics/claude-plugins-official under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.