Claude Code subagents
Specialised agents with their own instructions, tools and context window that the main Claude Code session can delegate tasks to.
- Part of
- Claude Code
- Defined in
- .claude/agents/*.md
Connected
- Claude CodeAnthropic's agentic coding tool that reads a codebase, edits files, runs commands and verifies its work from the terminal, IDE or web.
- Agent SkillsFolders of instructions, scripts and resources that Claude loads on demand when a task matches the skill's description, introduced by Anthropic in October 2025.
Linked from
- AI outside judgeAn acceptance check outside an agent’s self-assessment, such as a deterministic test, held-out evaluation, measurement or separately briefed reviewer.
- Multidisciplinary agent reviewA workflow in which specialist passes apply distinct methods and produce concrete evidence, rather than merely imitate expert personas.
More agent tooling & protocols
- agent-html-kitAn open-source template kit and Claude Code skill (kit.js) that lets agents emit content-only HTML components while design lives in one shared file, published with a 72-run HTML vs Markdown benchmark.
- CLAUDE.mdA Markdown file of standing instructions that Claude Code loads at the start of every session, at user, project and folder level.
- Karpathy-inspired CLAUDE.md rulesA widely starred public CLAUDE.md with four behaviour rules: think before coding, simplicity first, surgical changes and goal-driven execution.
- AGENTS.mdAn open, tool-neutral Markdown convention for giving coding agents project instructions, introduced by OpenAI in August 2025 and now stewarded by the Linux Foundation's Agentic AI Foundation.
- Claude Code hooksShell commands that Claude Code runs automatically at lifecycle events such as before or after a tool call, so rules can be enforced by code rather than requested in prose.
- Model Context Protocol (MCP)An open protocol, introduced by Anthropic in November 2024, that standardises how AI applications connect to external tools and data sources.