Software delivery workflows that let developers explore, plan, build, refactor, debug, test, and review code using orchestratable skills and multi-agent teams, with built-in TDD, PR generation, and QA pipelines.
Reference guide for Agent Teams orchestration mode. Defines how the /claude-setup:build and /claude-setup:refactor skills use Claude Code's native Agent Teams feature to implement tasks. NOT spawned as a sub-agent — the SKILL session executes these instructions directly.
Fast read-only project recon: discovers run/log/test commands and tech stack, writes .claude/app-context.md. Spawned by /claude-setup:debug-workflow and /claude-setup:qa.
Fixes a diagnosed bug using adaptive TDD: writes a failing test, implements the fix, and verifies no regressions. Reads $TASKS_DIR/bug-diagnosis.md. Spawned by /claude-setup:debug-workflow.
Investigates and diagnoses bugs by reading logs, tracing code, and reproducing issues. Produces $TASKS_DIR/bug-diagnosis.md with root cause, evidence, and fix recommendations. Supports DISCOVERY and DIAGNOSE modes. Spawned by /claude-setup:debug-workflow.
Reviews code changes against a PRD, task spec, or general quality standards. Produces a compliance report with Critical/Important/Minor issues. Spawned by /claude-setup:build and /claude-setup:debug-workflow.
Lightweight build: explore → plan → approve → implement in a single Opus context, no sub-agent fan-out. Verifies build + tests, optionally commits, and offers a separate /code-review pass at the end. Use by default for everyday features; reach for /claude-setup:build only when the work needs parallel fan-out or exceeds one context.
Full build pipeline: takes a PRD, breaks it into tasks, implements them in parallel, and reviews the result. Orchestrates prd-task-planner → parallel-task-orchestrator → code-reviewer.
Reads task files from tasks/<branch>/ and the diff against the detected base ref, then generates a polished PR description in markdown for the user to copy-paste.
Debug pipeline: investigates a bug, diagnoses root cause, writes failing tests, implements fix via TDD, and reviews the result. Orchestrates bug-investigator -> bug-fixer -> code-reviewer.
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when the user wants to stress-test a plan, get grilled on their design, or mentions "grill me". Run BEFORE /claude-setup:build when a feature idea is fuzzy and a PRD doesn't yet exist.
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A template for setting up Claude Code (agents, skills, memory) in any project. Use it to bootstrap a new repo or add Claude to an existing one.
What you get: a set of specialized agents and slash-command pipelines for common software tasks:
| Pipeline | What it does |
|---|---|
/grill-me, /grill-with-docs | Fuzzy idea → relentless interview → sharpened plan (run before /build) |
/build | PRD → cheap lite-routing → plan → parallel/direct implementation → diff-scoped code review |
/build-lite | Feature → plan → approve → implement in one context — no fan-out, review separated |
/debug-workflow | Bug report → investigate → diagnose → TDD fix → review |
/refactor | Target → audit → (write tests) → refactor → behavior-preservation review |
/refactor-lite | Target → audit → plan → approve → refactor in one context — no fan-out, review separated |
/qa | Running app → exploratory browser testing → QA report + Playwright E2E tests |
/craft-pr | Branch's task files + diff → polished PR description |
Task/QA artifacts are branch-scoped so you can run multiple pipelines in parallel across branches (including git worktrees) without collision.
Use this repo as a GitHub template to start a new project with Claude Code pre-configured:
git clone [email protected]:you/your-new-project.git
cd your-new-project
claude login # authenticate with your Max/Pro subscription
claude # start coding
Everything is ready out of the box — agents, skills, and settings are already in place. Add your project code and go.
By default the script installs agents and skills globally into ~/.claude/ — available across every project with no per-project setup needed:
bash <(curl -fsSL https://raw.githubusercontent.com/nickmaglowsch/claude-setup/main/setup.sh)
Re-run the same command to update. It auto-detects whether ~/.claude/agents/ already exists and updates in place (no prompts) or installs fresh (with overwrite prompts).
The template repo is auto-cloned to /tmp/claude-setup (or pulled if already there).
Use --local to install into a specific project directory instead:
cd /path/to/your/project
bash <(curl -fsSL https://raw.githubusercontent.com/nickmaglowsch/claude-setup/main/setup.sh) --local
The script auto-detects whether this is a first-time setup or an update:
.claude/agents/): runs the interactive setup — copies agents, skills, settings, and optionally adds devcontainer + headless runner.claude/agents/ found): updates all agent and skill files to the latest version while preserving your settings.local.json and agent-memory/Same one-liner with --local — the script auto-detects the existing setup and runs in update mode:
cd /path/to/your/project
bash <(curl -fsSL https://raw.githubusercontent.com/nickmaglowsch/claude-setup/main/setup.sh) --local
To force update mode explicitly:
bash <(curl -fsSL https://raw.githubusercontent.com/nickmaglowsch/claude-setup/main/setup.sh) --local --update
Update mode pulls the latest template, overwrites all agent and skill files, and leaves your settings.local.json and agent-memory/ untouched.
To add devcontainer support during an update (if you skipped it during initial setup):
bash <(curl -fsSL https://raw.githubusercontent.com/nickmaglowsch/claude-setup/main/setup.sh) --update --add-devcontainer
Add --compatible to generate native agent files for other tools alongside the Claude setup:
bash <(curl -fsSL https://raw.githubusercontent.com/nickmaglowsch/claude-setup/main/setup.sh) --compatible opencode,gemini,codex
For OpenCode and Gemini, you'll be prompted to choose models per agent system (heavy-tier and standard-tier). Generated files:
| Flag | Generated | Also creates |
|---|---|---|
opencode | .opencode/agents/*.md | AGENTS.md → CLAUDE.md symlink |
gemini | .gemini/agents/*.toml | GEMINI.md → CLAUDE.md symlink |
codex | ~/plugins/claude-setup-codex/ | ~/.agents/plugins/marketplace.json entry |
npx claudepluginhub nickmaglowsch/claude-setup --plugin claude-setupUniversal Claude Code workflow with specialized agents, skills, hooks, and output styles for any software project. Includes orchestrator, code-reviewer, debugger, docs-writer, security-auditor, refactorer, and test-architect agents.
Automation bootstrap for Claude Code. Analyzes any project and generates a complete .claude/ structure with agents, pipelines, skills, memory, hooks, and settings.
Software engineering workflows with skills for planning, implementation, quality review, and structured thinking, plus a suite of specialist agents
Complete collection of battle-tested Claude Code configs from an Anthropic hackathon winner - agents, skills, hooks, and rules evolved over 10+ months of intensive daily use
11 agents, 35 skills, 18 commands, 9 hooks — spec-driven multi-agent orchestration for Claude Code, with optional cross-device semantic memory.
Comprehensive skills library for Claude Code: planning, design, TDD, debugging, collaboration patterns, and proven techniques