claude-code-agents
Claude Code Agents Prompt templates for Claude Code's subagent system. Run parallel code audits, automate fix cycles, get stuff reviewed. Built for the Claude Code Task tool. Not a framework. Just prompts that work. What This Does Claude Code can spawn subagents via Task(). These are the prompts those agents receive.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/undeadlist--claude-code-agents)HTML
<a href="/listings/undeadlist--claude-code-agents"><img src="/api/listings/undeadlist--claude-code-agents/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
These steps run a script from the network that findsafeskills does not scan.
Install as Plugin (Recommended)
Install once, use across all projects. No per-project file copying needed.
# Step 1: Add the marketplace (inside Claude Code) /plugin marketplace add undeadlist/claude-code-agents # Step 2: Install the plugin (inside Claude Code) /plugin install claude-code-agents@undeadlist-claude-code-agents
Or after Step 1, use `/plugin` → **Discover** tab to browse and install.
You can also install from the terminal:
claude plugin install claude-code-agents@undeadlist-claude-code-agents --scope user
All 24 agents and 6 workflow skills become natively available. Use skills like `/full-audit`, `/pre-commit`, `/pre-deploy`, etc.
Verify Your Install
After installing, confirm everything loaded:
# Inside Claude Code, ask: "List all available agents and skills from the claude-code-agents plugin"
You should see 24 agents and 6 skills. If not, check the [Troubleshooting](#troubleshooting) section below.
Try Before Installing
Test the plugin locally without installing:
git clone https://github.com/undeadlist/claude-code-agents.git claude --plugin-dir ./claude-code-agents
This loads the plugin for that session only — nothing is installed globally.
Alternative Installation (Per-Project)
If you prefer per-project installation:
# Option A: Full setup (creates CLAUDE.md, detects stack) curl -s https://undeadlist.com/agents/setup-project.sh | bash # Option B: Just agents curl -s https://undeadlist.com/agents/install.sh | bash # Option C: Via npm npx claude-code-agents
This copies agents into your project's `.claude/agents/` directory.
---
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
.claude-plugin/marketplace.json · Manifest · 24 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
- {
- "name": "undeadlist-claude-code-agents",
- "owner": {
- "name": "UndeadList",
- "email": "hello@undeadlist.com"
- },
- "metadata": {
- "description": "24 parallel audit agents + 6 workflow skills for Claude Code",
- "version": "1.0.0"
- },
- "plugins": [
- {
- "name": "claude-code-agents",
- "source": "./",
- "description": "24 audit agents + 6 workflow skills for solo dev startups building with Next.js + Claude Code",
- "version": "3.1.0",
- "author": {
- "name": "UndeadList"
- },
- "category": "development-tools",
- "tags": ["audit", "security", "code-quality", "testing", "deployment", "qa", "solo-dev"]
- }
- ]
- }
README.md · README · 570 lines · 6 findings marked
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
- # Complete E2E Development Workflow
- [](https://undeadlist.com) [](https://github.com/hesreallyhim/awesome-claude-code)
- ## For Solo Dev Startups Building with Next.js + Claude Code
- 
- > **Target User:** Solo dev running a startup, only engineer, entire company.
- > **Goal:** Pull this repo into any project and have a full agent team ready to go.
- > **Designed for:** Next.js / React / TypeScript full-stack web apps (Prisma, npm/pnpm, Vercel).
- > Other stacks can adapt the agent prompts, but examples and tooling defaults target this stack.
- > **Built by:** Paul @ UndeadList — learned the hard way what works.
- ---
- ## The Reality Check
- You're a solo dev. You don't have:
- - A QA team
- - A code reviewer
- - A DevOps engineer
- - Time to waste on AI agents going rogue
- This workflow package gives you all of that via Claude Code subagents, with strict protocols to prevent the bullshit that wastes your time.
- ---
- ## What's In This Repo
- ```
- claude-code-agents/
- ├── .claude-plugin/
- │ ├── plugin.json # Plugin manifest
- │ └── marketplace.json # Marketplace config
- │
- ├── agents/ # 24 agent definitions
- │ ├── # AUDIT AGENTS (11 - run in parallel)
- │ ├── code-auditor.md # Code quality, DRY, complexity
- │ ├── bug-auditor.md # Runtime bugs, auth gaps
- │ ├── security-auditor.md # OWASP deep scan
- │ ├── doc-auditor.md # Documentation gaps
- │ ├── infra-auditor.md # Config, env vars, headers
- │ ├── ui-auditor.md # Accessibility, UX
- │ ├── db-auditor.md # Database, N+1, indexes
- │ ├── perf-auditor.md # Performance, bundle size
- │ ├── dep-auditor.md # Dependencies, vulnerabilities
- │ ├── seo-auditor.md # SEO, meta tags, OpenGraph
- │ ├── api-tester.md # API endpoint testing
- │ ├── # FIX/IMPLEMENT AGENTS (4)
- │ ├── fix-planner.md # Prioritizes findings into FIXES.md
- │ ├── code-fixer.md # Implements fixes
- │ ├── test-runner.md # Validates fixes
- │ ├── test-writer.md # Auto-generates tests
- │ ├── # BROWSER AGENTS (4 - Chrome integration)
- │ ├── browser-qa-agent.md # Navigates UI, finds console errors
- │ ├── fullstack-qa-orchestrator.md # Find → Fix → Verify loop
- │ ├── console-monitor.md # Real-time console watching
- │ ├── visual-diff.md # Screenshot comparison
- │ ├── # DEPLOY AGENTS (2)
- │ ├── deploy-checker.md # Pre-deployment validation
- │ ├── env-validator.md # Environment configuration
- │ ├── # UTILITY AGENTS (2)
- │ ├── pr-writer.md # PR description generator
- │ ├── seed-generator.md # Test data creation
- │ └── # SUPERVISORS (1)
- │ └── architect-reviewer.md # Oversees until production-ready
- │
- ├── skills/ # 6 workflow skills (slash commands)
- │ ├── full-audit/SKILL.md # /full-audit
- │ ├── pre-commit/SKILL.md # /pre-commit
- │ ├── pre-deploy/SKILL.md # /pre-deploy
- │ ├── new-feature/SKILL.md # /new-feature
- │ ├── bug-fix/SKILL.md # /bug-fix
- │ └── release-prep/SKILL.md # /release-prep
- │
- ├── workflows/ # Detailed workflow documentation
- │ ├── pre-commit.md
- │ ├── pre-deploy.md
- │ ├── full-audit.md
- │ ├── new-feature.md
- │ ├── bug-fix.md
- │ └── release-prep.md
- │
- ├── CLAUDE.md # Orchestrator instructions
- ├── setup-project.sh # One command setup for new projects
- ├── install.sh # Quick agent install (legacy)
- └── README.md
- ```
- **Total: 24 agents + 6 workflow skills**
- ---
- ## Quick Start
- ### Install as Plugin (Recommended)
- Install once, use across all projects. No per-project file copying needed.
- ```bash
- # Step 1: Add the marketplace (inside Claude Code)
- /plugin marketplace add undeadlist/claude-code-agents
- # Step 2: Install the plugin (inside Claude Code)
- /plugin install claude-code-agents@undeadlist-claude-code-agents
- ```
- Or after Step 1, use `/plugin` → **Discover** tab to browse and install.
- You can also install from the terminal:
- ```bash
- claude plugin install claude-code-agents@undeadlist-claude-code-agents --scope user
- ```
- All 24 agents and 6 workflow skills become natively available. Use skills like `/full-audit`, `/pre-commit`, `/pre-deploy`, etc.
- ### Verify Your Install
- After installing, confirm everything loaded:
- ```bash
- # Inside Claude Code, ask:
- "List all available agents and skills from the claude-code-agents plugin"
- ```
- You should see 24 agents and 6 skills. If not, check the [Troubleshooting](#troubleshooting) section below.
- ### Try Before Installing
- Test the plugin locally without installing:
- ```bash
- git clone https://github.com/undeadlist/claude-code-agents.git
- claude --plugin-dir ./claude-code-agents
- ```
- This loads the plugin for that session only — nothing is installed globally.
- ### Alternative Installation (Per-Project)
- If you prefer per-project installation:
- ```bash
- # Option A: Full setup (creates CLAUDE.md, detects stack)
- curl -s https://undeadlist.com/agents/setup-project.sh | bash
- Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
- Line 146 on GitHub
- # Option B: Just agents
- curl -s https://undeadlist.com/agents/install.sh | bash
- Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
- Line 149 on GitHub
- # Option C: Via npm
- npx claude-code-agents
- ```
- This copies agents into your project's `.claude/agents/` directory.
- ---
- ## Stack Compatibility
- These agents are optimized for **Next.js / React / TypeScript** projects using **Prisma**, **npm/pnpm**, and **Vercel**. Commands, file paths, and examples throughout the agent prompts reflect this stack.
- **Stack-agnostic agents** (useful for any project):
- - `security-auditor` — OWASP checks apply to any web app
- - `code-auditor` — quality principles are language-agnostic
- - `doc-auditor` — documentation gap analysis works everywhere
- - `pr-writer` — generates PR descriptions for any git repo
- - `fix-planner` — prioritization logic is framework-independent
- - `architect-reviewer` — review criteria are generic
- **Adapting for other stacks:** Fork the repo and edit the agent `.md` files in `agents/`. Swap Next.js-specific commands (`npm run dev`, `pnpm tsc`, `localhost:3000`) for your equivalents. The orchestration structure works regardless of stack — only the tool-specific examples need changing.
- ---
- ## Workflows
- ### Pre-Commit
- ```
- claude "Run pre-commit workflow"
- ```
- Runs code-auditor + test-runner before committing.
- ### Pre-Deploy
- ```
- claude "Run pre-deploy workflow"
- ```
- Validates build, env vars, dependencies before deployment.
- ### Full Audit
- ```
- claude "Run full-audit workflow on src/"
- ```
- All 11 auditors in parallel → fix-planner creates FIXES.md.
- ### New Feature
- ```
- claude "Run new-feature workflow for: [feature description]"
- ```
- test-writer → code-fixer → test-runner → browser-qa (TDD approach).
- ### Bug Fix
- ```
- claude "Run bug-fix workflow for: [bug description]"
- ```
- Write failing test → fix → verify (regression prevention).
- ### Release Prep
- ```
- claude "Run release-prep workflow for v1.0.0"
- ```
- Full audit → fixes → deploy validation → PR generation.
- ---
- ## Project Lifecycle
- ### Phase 0: New Project Setup
- ```bash
- curl -s https://undeadlist.com/agents/setup-project.sh | bash
- Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
- Line 220 on GitHub
- ```
- ### Phase 1: Skeleton/Scaffolding
- **Before touching code, create your blueprint:**
- ```
- claude "Create a SETUP_BLUEPRINT.md for this project:
- - Folder structure
- - Tech stack decisions (with rationale)
- - Dependencies list (don't install yet)
- - Environment variables needed
- - Database planning
- - Success criteria checklist
- Do NOT generate any code. Just the plan."
- ```
- ### Phase 2: Initial Build
- **Strict rules for AI agents during build:**
- ```markdown
- ## RULES FOR THIS SESSION
- 1. **ASK BEFORE** modifying any file not directly related to current task
- 2. **NEVER** refactor "while you're in there"
- 3. **NEVER** update dependencies unless explicitly requested
- 4. **NEVER** change config files without approval
- 5. **ONE logical change per commit**
- 6. **VERIFY** the build works after each change
- ```
- ### Phase 3: Development Loop
- The core loop for any task:
- ```
- ┌─────────────────────────────────────────────────┐
- │ 1. Understand the task │
- │ 2. State what files you'll touch │
- │ 3. Get approval │
- │ 4. Make ONE change │
- │ 5. Verify it works │
- │ 6. Commit │
- │ 7. Repeat │
- └─────────────────────────────────────────────────┘
- ```
- ### Phase 4: Pre-Launch Audit
- Run the full parallel audit:
- ```
- claude "Run full-audit workflow on src/"
- ```
- **Review `.claude/audits/` for:**
- - AUDIT_CODE.md, AUDIT_SECURITY.md, AUDIT_DB.md, AUDIT_PERF.md
- - FIXES.md (prioritized action items)
- ### Phase 5: Browser QA
- Enable Chrome integration:
- ```
- /chrome
- ```
- Run the full QA loop:
- ```
- claude "Use fullstack-qa-orchestrator to:
- 1. Start dev server (npm run dev)
- 2. Navigate to http://localhost:3000
- 3. Test: homepage, auth flow, main features
- 4. Find console errors, UI bugs
- 5. Fix issues found
- 6. Re-test to verify
- 7. Repeat until clean"
- ```
- ### Phase 6: Deploy
- ```
- claude "Run pre-deploy workflow"
- ```
- ---
- ## The Protocols
- ### Protocol 1: No Unauthorized Changes
- Every AI agent must follow:
- ```markdown
- ## BEFORE ANY CHANGE
- 1. State what file you're modifying
- 2. State what change you're making
- 3. State why
- 4. Wait for "PROCEED" or "STOP"
- ## FORBIDDEN ACTIONS
- - ❌ npm install / npx anything without approval
- - ❌ Modifying package.json without approval
- - ❌ Creating new files when editing existing works
- - ❌ Refactoring unrelated code
- - ❌ "Improvements" not requested
- - ❌ Touching config files without approval
- ```
- ### Protocol 2: Micro-Checkpoint Method
- For complex changes:
- ```markdown
- ## EVERY PHASE:
- 1. Agent states planned change
- 2. Agent calls architect-reviewer for verification
- 3. User says "PROCEED" or "STOP"
- 4. Agent makes ONE change
- 5. Agent verifies change applied correctly
- 6. Proceed to next phase only with approval
- ```
- ### Protocol 3: Regression Prevention
- Before any change:
- ```bash
- # Baseline testing
- curl -s http://localhost:3000/api/health # API works
- pnpm tsc --noEmit # TypeScript valid
- pnpm test # Tests pass
- # After change - all must still work
- ```
- ---
- ## Common Commands
- ### Daily Development
- ```bash
- # Start coding session
- claude "Review CLAUDE.md, then help me with [task]"
- # Quick check before commit
- claude "Run pre-commit workflow"
- # Verify a specific file
- claude "Review [file] for bugs, especially [concern area]"
- ```
- ### Pre-Launch
- ```bash
- # Full audit
- claude "Run full-audit workflow on src/"
- # Browser QA
- claude "Use fullstack-qa-orchestrator to test http://localhost:3000"
- # Security focus
- claude "Use security-auditor to deep scan src/api/"
- ```
- ### Debugging
- ```bash
- # Console monitoring
- claude "Use console-monitor to watch http://localhost:3000"
- # Specific bug
- claude "Run bug-fix workflow for: [bug description]"
- ```
- ---
- ## Files Generated
- All reports go to `.claude/audits/` (gitignored):
- | File | Source Agent |
- |------|--------------|
- | AUDIT_CODE.md | code-auditor |
- | AUDIT_SECURITY.md | bug-auditor |
- | AUDIT_SECURITY_DEEP.md | security-auditor |
- | AUDIT_DOCS.md | doc-auditor |
- | AUDIT_INFRA.md | infra-auditor |
- | AUDIT_UI_UX.md | ui-auditor |
- | AUDIT_DB.md | db-auditor |
- | AUDIT_PERF.md | perf-auditor |
- | AUDIT_DEPS.md | dep-auditor |
- | AUDIT_SEO.md | seo-auditor |
- | API_TEST_REPORT.md | api-tester |
- | DEPLOY_CHECK.md | deploy-checker |
- | ENV_REPORT.md | env-validator |
- | FIXES.md | fix-planner |
- | TEST_REPORT.md | test-runner |
- | AUDIT_BROWSER_QA.md | browser-qa-agent |
- | QA_SESSION_LOG.md | fullstack-qa-orchestrator |
- ---
- ## Quick Reference
- | Task | Command |
- |------|---------|
- | New project setup | `curl -s https://undeadlist.com/agents/setup-project.sh \| bash` |
- Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
- Line 433 on GitHub
- | Enable browser QA | `/chrome` in Claude Code |
- | Pre-commit check | `"Run pre-commit workflow"` |
- | Full audit | `"Run full-audit workflow on src/"` |
- | Pre-deploy check | `"Run pre-deploy workflow"` |
- | Browser test | `"Use browser-qa-agent to test localhost:3000"` |
- | Full QA loop | `"Use fullstack-qa-orchestrator..."` |
- | New feature | `"Run new-feature workflow for: [feature]"` |
- | Bug fix | `"Run bug-fix workflow for: [bug]"` |
- | Release prep | `"Run release-prep workflow for v1.0.0"` |
- | Verify production-ready | `"Use architect-reviewer..."` |
- ---
- ## When AI Goes Rogue
- ### Red Flags - STOP Immediately
- - Agent makes changes without asking
- - Agent installs packages without approval
- - Agent modifies config files unexpectedly
- - Agent claims success without verification
- - Agent skips checkpoints
- - Agent touches files outside scope
- ### Recovery Commands
- ```bash
- # Revert last change
- git checkout -- [file]
- # Revert all changes
- git reset --hard HEAD
- # Check what changed
- git diff
- ```
- ### Enforcement Phrases
- ```
- "STOP. You didn't ask permission."
- "STOP. That's not in scope."
- "STOP. Verify it works before claiming done."
- "STOP. One change at a time."
- ```
- ---
- ## Troubleshooting
- ### Plugin not loading / no agents found
- 1. **Check plugin errors:** Run `/plugin` inside Claude Code and check the **Errors** tab
- 2. **Verify marketplace was added:** Go to `/plugin` → **Marketplaces** tab — you should see `undeadlist-claude-code-agents`
- 3. **Verify plugin is installed:** Go to `/plugin` → **Installed** tab — you should see `claude-code-agents`
- 4. **Re-add and reinstall:**
- ```bash
- /plugin marketplace add undeadlist/claude-code-agents
- /plugin install claude-code-agents@undeadlist-claude-code-agents
- ```
- ### Skills (slash commands) not appearing
- Skills like `/full-audit` only work when installed as a plugin. Per-project installs (`npx claude-code-agents`) install agents to `.claude/agents/` but don't register slash commands — use natural language instead:
- ```
- claude "Run full-audit workflow on src/"
- ```
- ### `npx claude-code-agents` fails
- - Requires Node.js >= 14
- - Must be run from a project directory
- - Copies agents to `.claude/agents/` and workflows to `workflows/` in your project
- ### `curl` installer returns 404
- The setup script URL is `https://undeadlist.com/agents/setup-project.sh` (not `setup.sh`). If the CDN isn't configured, use the GitHub raw URL:
- ```bash
- curl -sL https://raw.githubusercontent.com/undeadlist/claude-code-agents/main/install.sh | bash
- Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
- Line 512 on GitHub
- ```
- ### Agents exist but aren't being used
- Make sure your project has a `CLAUDE.md` that references the agents. Run `setup-project.sh` to auto-generate one, or see the [CLAUDE.md](CLAUDE.md) in this repo as a template.
- ---
- ## Uninstall
- ### Plugin
- ```bash
- # In Claude Code:
- /plugin uninstall claude-code-agents
- ```
- ### Per-Project Install
- ```bash
- # Option A: One-liner (recommended)
- curl -s https://raw.githubusercontent.com/undeadlist/claude-code-agents/main/uninstall.sh | bash
- Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
- Line 532 on GitHub
- # Option B: Manual removal
- rm -rf .claude/agents/ .claude/audits/ workflows/
- ```
- This removes:
- - `.claude/agents/` - agent prompt definitions
- - `.claude/audits/` - generated reports (gitignored)
- - `workflows/` - predefined workflow chains
- > **Note:** CLAUDE.md is left in place. Remove manually if desired.
- ---
- ## Contributing
- Found a better protocol? Agent that should exist?
- PR it: https://github.com/undeadlist/claude-code-agents
- ---
- ## License
- MIT. Use it, fork it, enhance it.
- ---
- **Built by [UndeadList](https://undeadlist.com)**
- *The indie software flea market*
- ---
- **Disclaimer:** Claude Code is a product of [Anthropic PBC](https://anthropic.com). This project is not affiliated with or endorsed by Anthropic.
- ---
- *"Built, but undiscovered."*
install.sh · Bundled script · 111 lines · 1 finding marked
- #!/bin/bash
- # claude-code-agents installer
- # Usage: curl -s https://undeadlist.com/agents/install.sh | bash
- Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
- Line 4 on GitHub
- set -e
- REPO_URL="https://github.com/undeadlist/claude-code-agents"
- BRANCH="main"
- echo "claude-code-agents installer"
- echo "================================"
- echo ""
- # Check if we're in a project directory
- if [ ! -f "package.json" ] && [ ! -f "Cargo.toml" ] && [ ! -f "go.mod" ] && [ ! -f "requirements.txt" ]; then
- echo "Warning: No project file detected (package.json, Cargo.toml, etc.)"
- echo " Are you in your project root?"
- read -p " Continue anyway? [y/N] " -n 1 -r
- echo
- if [[ ! $REPLY =~ ^[Yy]$ ]]; then
- exit 1
- fi
- fi
- # Create directories
- echo "Creating directories..."
- mkdir -p .claude/agents
- mkdir -p .claude/audits
- mkdir -p workflows
- # Download agent files (24 total)
- echo "Downloading agents..."
- AGENTS=(
- # Audit agents (11)
- "code-auditor.md"
- "bug-auditor.md"
- "security-auditor.md"
- "doc-auditor.md"
- "infra-auditor.md"
- "ui-auditor.md"
- "db-auditor.md"
- "perf-auditor.md"
- "dep-auditor.md"
- "seo-auditor.md"
- "api-tester.md"
- # Fix/Implement agents (4)
- "fix-planner.md"
- "code-fixer.md"
- "test-runner.md"
- "test-writer.md"
- # Browser QA agents (4)
- "browser-qa-agent.md"
- "fullstack-qa-orchestrator.md"
- "console-monitor.md"
- "visual-diff.md"
- # Deploy agents (2)
- "deploy-checker.md"
- "env-validator.md"
- # Utility agents (2)
- "pr-writer.md"
- "seed-generator.md"
- # Supervisors (1)
- "architect-reviewer.md"
- )
- for agent in "${AGENTS[@]}"; do
- echo " -> $agent"
- curl -sL "${REPO_URL}/raw/${BRANCH}/agents/${agent}" -o ".claude/agents/${agent}"
- done
- # Download workflow files (6 total)
- echo ""
- echo "Downloading workflows..."
- WORKFLOWS=(
- "pre-commit.md"
- "pre-deploy.md"
- "full-audit.md"
- "new-feature.md"
- "bug-fix.md"
- "release-prep.md"
- )
- for workflow in "${WORKFLOWS[@]}"; do
- echo " -> $workflow"
- curl -sL "${REPO_URL}/raw/${BRANCH}/workflows/${workflow}" -o "workflows/${workflow}"
- done
- # Create .gitignore for audits
- if [ ! -f ".claude/.gitignore" ]; then
- echo "audits/" > .claude/.gitignore
- fi
- echo ""
- echo "================================"
- echo "Installed ${#AGENTS[@]} agents to .claude/agents/"
- echo "Installed ${#WORKFLOWS[@]} workflows to workflows/"
- echo ""
- echo "Quick start:"
- echo " claude \"Run full-audit workflow on src/\""
- echo " claude \"Run pre-commit workflow\""
- echo " claude \"Run pre-deploy workflow\""
- echo ""
- echo "Tip: Install as a plugin instead (once, works across all projects):"
- echo " /plugin marketplace add undeadlist/claude-code-agents"
- echo " claude plugin install claude-code-agents@undeadlist-claude-code-agents"
- echo ""
- echo "Docs: https://github.com/undeadlist/claude-code-agents"
- echo "Built by UndeadList: https://undeadlist.com"
uninstall.sh · Bundled script · 48 lines · 1 finding marked
- #!/bin/bash
- # claude-code-agents uninstaller
- # Usage: curl -s https://raw.githubusercontent.com/undeadlist/claude-code-agents/main/uninstall.sh | bash
- Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
- Line 3 on GitHub
- set -e
- echo "claude-code-agents uninstaller"
- echo "================================"
- echo ""
- removed=0
- # Remove agent definitions
- if [ -d ".claude/agents" ]; then
- rm -rf .claude/agents/
- echo "Removed .claude/agents/"
- ((removed++))
- fi
- # Remove generated reports
- if [ -d ".claude/audits" ]; then
- rm -rf .claude/audits/
- echo "Removed .claude/audits/"
- ((removed++))
- fi
- # Remove workflows
- if [ -d "workflows" ]; then
- rm -rf workflows/
- echo "Removed workflows/"
- ((removed++))
- fi
- # Clean up empty .claude directory if nothing else is in it
- if [ -d ".claude" ] && [ -z "$(ls -A .claude 2>/dev/null)" ]; then
- rmdir .claude
- echo "Removed empty .claude/"
- fi
- echo ""
- if [ $removed -eq 0 ]; then
- echo "Nothing to remove. Are you in the right project directory?"
- else
- echo "================================"
- echo "Uninstall complete."
- echo ""
- echo "Note: CLAUDE.md was left in place (remove manually if desired)."
- fi
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.