Claude Code Setup

Official Anthropic plugin that scans a repo and recommends the top 1-2 Claude Code automations per category — MCP servers, skills, hooks, subagents, slash commands — tailored to the detected stack. Read-only: it advises, it does not modify files. Source: anthropics/claude-plugins-official, https://github.com/anthropics/claude-plugins-official/tree/main/plugins/claude-code-setup, 2026-06-12

Most developers run Claude Code "vanilla" - as a glorified terminal chat - and miss the ecosystem (CLAUDE.md + hooks + skills + subagents + MCP) that turns it into a real dev environment. claude-code-setup closes that gap by analyzing the actual project and naming the highest-impact configuration for that codebase, instead of leaving the user to guess. It ships inside Anthropic's official plugin directory (anthropics/claude-plugins-official, 31,476 stars at the 2026-07-03 source check) and is authored by Isabella He of Anthropic. Source: anthropics/claude-plugins-official README, 2026-07-03

What it recommends

The plugin's claude-automation-recommender skill (tools: Read, Glob, Grep, Bash) detects the stack, then proposes the top one or two automations in each category: Source: claude-automation-recommender SKILL.md, GitHub, 2026-06-12

Category What it adds Example recommendations
MCP servers External integrations context7 for docs, Playwright for frontend
Skills Packaged expertise / workflows Plan agent, frontend-design, feature-dev
Hooks Automatic actions on tool events auto-format/auto-lint on edit, block .env and lock-file edits, run related tests
Subagents Specialized parallel reviewers security, performance, accessibility
Slash commands One-liner workflows /test, /pr-review, /explain

It detects framework + tooling signals (e.g. Poetry/FastAPI/Pydantic, ESLint/Prettier, Tailwind, CI) and maps each to a guarded recommendation — Prettier present → PostToolUse format-on-edit; .env present → PreToolUse block. Source: Towards AI / AlphaSignal walkthroughs, 2026-05

Install and use

/plugin install claude-code-setup@claude-plugins-official
# then, inside the project:
"recommend automations for this project"   # or "help me set up Claude Code" / "what hooks should I use?"

The plugin is read-only — it produces a recommendation report; you accept, reject, or ask for more options per category, then implement the suggested hooks/skills yourself. Source: anthropics/claude-plugins-official README, 2026-06-12

Version Snapshot

Checked 2026-07-03:

Surface Current
GitHub anthropics/claude-plugins-official, Apache-2.0, 31,476 stars, 3,444 forks
Source revision main HEAD 358ee646f45b39424320d47abbb15554d4eb31e3; no public tags found in the checked ref query
Plugin path plugins/claude-code-setup/ with .claude-plugin, README, LICENSE, automation-recommender-example.png, and skills/claude-automation-recommender/
Installation surface /plugin install claude-code-setup@claude-plugins-official in Claude Code

The reviewed X video shows the practical install UX: Claude Code v2.1.119 running /plugin install claude-code-setup@claude-plugins-official, plugin discovery/details, user-scope versus project-scope install choices, and the follow-up instruction to run /reload-plugins. The local thumbnail only captures the install command, so the video frames are the source for those UI observations. Source: X/@Suryanshti777, 2026-05-17; Source: local video contact sheet, 2026-07-03

Where it fits Kevin's stack

It is the official, generic analog of Kevin's bespoke harness work: the Agent-Docs Mesh enrolls repos with AGENTS.md + per-dir SKILL.md, and Claude Code Harness / Everything Claude Code - Agent Harness Performance System document the same hooks-skills-subagents-MCP surface by hand. claude-code-setup is useful as a first-pass audit on an unfamiliar repo before deeper enrollment — its recommendation step parallels the doctor/skills-first habit, and its "top 1-2 per category" restraint matches Kevin's quality-over-quantity stance on skills. It pairs naturally with the Claude Code /goal Command /goal loop (set up the environment, then run goal-directed sessions against it).


Timeline

  • 2026-07-03 | Deep-reviewed the bookmark video and refreshed official source authority: anthropics/claude-plugins-official, Apache-2.0, 31,476 stars, 3,444 forks, main HEAD 358ee646f45b39424320d47abbb15554d4eb31e3, no public tags found. The video shows Claude Code v2.1.119 installing claude-code-setup@claude-plugins-official, plugin details, install-scope choices, and /reload-plugins. Source: GitHub API/ref check and X media review, 2026-07-03
  • 2026-05-17 | @Suryanshti777 surfaced the plugin ("Anthropic quietly released an official plugin called claude-code-setup… turns Claude Code from 'pretty good' into an actual AI dev environment. It scans your project and recommends: hooks…"). 4,564 likes, 13,851 bookmarks, demo video. Source: X/@Suryanshti777, 2026-05-17
  • 2026-06-12 | Page created from coding-agents bookmark absorption; promoted from a subsection of Claude Code /goal Command to a dedicated tool page. Verified against the official anthropics/claude-plugins-official repo (author: Isabella He, Anthropic; read-only recommender). Source: anthropics/claude-plugins-official, 2026-06-12