{"id":44394780,"url":"https://github.com/undeadlist/claude-code-agents","last_synced_at":"2026-02-13T19:01:05.169Z","repository":{"id":330497730,"uuid":"1122137539","full_name":"undeadlist/claude-code-agents","owner":"undeadlist","description":"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.","archived":false,"fork":false,"pushed_at":"2025-12-26T05:10:16.000Z","size":100,"stargazers_count":4,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-12-27T08:59:18.901Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Shell","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/undeadlist.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-12-24T06:55:42.000Z","updated_at":"2025-12-26T05:17:17.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/undeadlist/claude-code-agents","commit_stats":null,"previous_names":["undeadlist/claude-code-agents"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/undeadlist/claude-code-agents","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/undeadlist%2Fclaude-code-agents","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/undeadlist%2Fclaude-code-agents/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/undeadlist%2Fclaude-code-agents/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/undeadlist%2Fclaude-code-agents/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/undeadlist","download_url":"https://codeload.github.com/undeadlist/claude-code-agents/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/undeadlist%2Fclaude-code-agents/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29414343,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-02-13T06:24:03.484Z","status":"ssl_error","status_checked_at":"2026-02-13T06:23:12.830Z","response_time":78,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2026-02-12T03:01:10.128Z","updated_at":"2026-02-13T19:01:05.154Z","avatar_url":"https://github.com/undeadlist.png","language":"Shell","funding_links":[],"categories":["Agent Skills 🤖","Web \u0026 Search","代理技能 🤖"],"sub_categories":["General","通用"],"readme":"# Complete E2E Development Workflow\n\n[![Claude Code Ready](badges/claude-code-ready.svg)](https://undeadlist.com)\n\n## For Solo Dev Startups Using Claude Code Agents\n\n![Parallel Audit Demo](./assets/terminal-demo.svg)\n\n\u003e **Target User:** Solo dev running a startup, only engineer, entire company.\n\u003e **Goal:** Pull this repo into any project and have a full agent team ready to go.\n\u003e **Built by:** Paul @ UndeadList — learned the hard way what works.\n\n---\n\n## The Reality Check\n\nYou're a solo dev. You don't have:\n- A QA team\n- A code reviewer\n- A DevOps engineer\n- Time to waste on AI agents going rogue\n\nThis workflow package gives you all of that via Claude Code subagents, with strict protocols to prevent the bullshit that wastes your time.\n\n---\n\n## What's In This Repo\n\n```\nclaude-code-agents/\n├── .claude/\n│   └── agents/\n│       ├── # AUDIT AGENTS (11 - run in parallel)\n│       ├── code-auditor.md         # Code quality, DRY, complexity\n│       ├── bug-auditor.md          # Security vulns, auth gaps\n│       ├── security-auditor.md     # OWASP deep scan\n│       ├── doc-auditor.md          # Documentation gaps\n│       ├── infra-auditor.md        # Config, env vars, headers\n│       ├── ui-auditor.md           # Accessibility, UX\n│       ├── db-auditor.md           # Database, N+1, indexes\n│       ├── perf-auditor.md         # Performance, bundle size\n│       ├── dep-auditor.md          # Dependencies, vulnerabilities\n│       ├── seo-auditor.md          # SEO, meta tags, OpenGraph\n│       ├── api-tester.md           # API endpoint testing\n│       │\n│       ├── # FIX/IMPLEMENT AGENTS (4)\n│       ├── fix-planner.md          # Prioritizes findings into FIXES.md\n│       ├── code-fixer.md           # Implements fixes\n│       ├── test-runner.md          # Validates fixes\n│       ├── test-writer.md          # Auto-generates tests\n│       │\n│       ├── # BROWSER AGENTS (4 - Chrome integration)\n│       ├── browser-qa-agent.md     # Navigates UI, finds console errors\n│       ├── fullstack-qa-orchestrator.md  # Find → Fix → Verify loop\n│       ├── console-monitor.md      # Real-time console watching\n│       ├── visual-diff.md          # Screenshot comparison\n│       │\n│       ├── # DEPLOY AGENTS (2)\n│       ├── deploy-checker.md       # Pre-deployment validation\n│       ├── env-validator.md        # Environment configuration\n│       │\n│       ├── # UTILITY AGENTS (2)\n│       ├── pr-writer.md            # PR description generator\n│       ├── seed-generator.md       # Test data creation\n│       │\n│       └── # SUPERVISORS (1)\n│       └── architect-reviewer.md   # Oversees until production-ready\n│\n├── workflows/                      # Predefined agent chains\n│   ├── pre-commit.md               # Before committing\n│   ├── pre-deploy.md               # Before deployment\n│   ├── full-audit.md               # Complete audit suite\n│   ├── new-feature.md              # Test-first development\n│   ├── bug-fix.md                  # Bug fix with regression prevention\n│   └── release-prep.md             # Release preparation\n│\n├── CLAUDE.md.template              # Project config (customize per project)\n├── setup-project.sh                # One command setup for new projects\n├── install.sh                      # Quick agent install\n└── README.md\n```\n\n**Total: 24 agents + 6 workflows**\n\n---\n\n## Quick Start\n\n### Installation\n\n```bash\n# Option A: Full setup (creates CLAUDE.md, detects stack)\ncurl -s https://undeadlist.com/agents/setup.sh | bash\n\n# Option B: Just agents\ncurl -s https://undeadlist.com/agents/install.sh | bash\n\n# Option C: Via npm\nnpx claude-code-agents\n```\n\nThis creates:\n- `.claude/agents/` with all 24 agent definitions\n- `.claude/audits/` for generated reports (gitignored)\n- `CLAUDE.md` with your project config\n- `workflows/` with predefined agent chains\n\n---\n\n## Workflows\n\n### Pre-Commit\n```\nclaude \"Run pre-commit workflow\"\n```\nRuns code-auditor + test-runner before committing.\n\n### Pre-Deploy\n```\nclaude \"Run pre-deploy workflow\"\n```\nValidates build, env vars, dependencies before deployment.\n\n### Full Audit\n```\nclaude \"Run full-audit workflow on src/\"\n```\nAll 11 auditors in parallel → fix-planner creates FIXES.md.\n\n### New Feature\n```\nclaude \"Run new-feature workflow for: [feature description]\"\n```\ntest-writer → code-fixer → test-runner → browser-qa (TDD approach).\n\n### Bug Fix\n```\nclaude \"Run bug-fix workflow for: [bug description]\"\n```\nWrite failing test → fix → verify (regression prevention).\n\n### Release Prep\n```\nclaude \"Run release-prep workflow for v1.0.0\"\n```\nFull audit → fixes → deploy validation → PR generation.\n\n---\n\n## Project Lifecycle\n\n### Phase 0: New Project Setup\n\n```bash\ncurl -s https://undeadlist.com/agents/setup.sh | bash\n```\n\n### Phase 1: Skeleton/Scaffolding\n\n**Before touching code, create your blueprint:**\n\n```\nclaude \"Create a SETUP_BLUEPRINT.md for this project:\n- Folder structure\n- Tech stack decisions (with rationale)\n- Dependencies list (don't install yet)\n- Environment variables needed\n- Database planning\n- Success criteria checklist\n\nDo NOT generate any code. Just the plan.\"\n```\n\n### Phase 2: Initial Build\n\n**Strict rules for AI agents during build:**\n\n```markdown\n## RULES FOR THIS SESSION\n\n1. **ASK BEFORE** modifying any file not directly related to current task\n2. **NEVER** refactor \"while you're in there\"\n3. **NEVER** update dependencies unless explicitly requested\n4. **NEVER** change config files without approval\n5. **ONE logical change per commit**\n6. **VERIFY** the build works after each change\n```\n\n### Phase 3: Development Loop\n\nThe core loop for any task:\n\n```\n┌─────────────────────────────────────────────────┐\n│  1. Understand the task                         │\n│  2. State what files you'll touch               │\n│  3. Get approval                                │\n│  4. Make ONE change                             │\n│  5. Verify it works                             │\n│  6. Commit                                      │\n│  7. Repeat                                      │\n└─────────────────────────────────────────────────┘\n```\n\n### Phase 4: Pre-Launch Audit\n\nRun the full parallel audit:\n\n```\nclaude \"Run full-audit workflow on src/\"\n```\n\n**Review `.claude/audits/` for:**\n- AUDIT_CODE.md, AUDIT_SECURITY.md, AUDIT_DB.md, AUDIT_PERF.md\n- FIXES.md (prioritized action items)\n\n### Phase 5: Browser QA\n\nEnable Chrome integration:\n```\n/chrome\n```\n\nRun the full QA loop:\n```\nclaude \"Use fullstack-qa-orchestrator to:\n1. Start dev server (npm run dev)\n2. Navigate to http://localhost:3000\n3. Test: homepage, auth flow, main features\n4. Find console errors, UI bugs\n5. Fix issues found\n6. Re-test to verify\n7. Repeat until clean\"\n```\n\n### Phase 6: Deploy\n\n```\nclaude \"Run pre-deploy workflow\"\n```\n\n---\n\n## The Protocols\n\n### Protocol 1: No Unauthorized Changes\n\nEvery AI agent must follow:\n\n```markdown\n## BEFORE ANY CHANGE\n\n1. State what file you're modifying\n2. State what change you're making\n3. State why\n4. Wait for \"PROCEED\" or \"STOP\"\n\n## FORBIDDEN ACTIONS\n\n- ❌ npm install / npx anything without approval\n- ❌ Modifying package.json without approval\n- ❌ Creating new files when editing existing works\n- ❌ Refactoring unrelated code\n- ❌ \"Improvements\" not requested\n- ❌ Touching config files without approval\n```\n\n### Protocol 2: Micro-Checkpoint Method\n\nFor complex changes:\n\n```markdown\n## EVERY PHASE:\n\n1. Agent states planned change\n2. Agent calls architect-reviewer for verification\n3. User says \"PROCEED\" or \"STOP\"\n4. Agent makes ONE change\n5. Agent verifies change applied correctly\n6. Proceed to next phase only with approval\n```\n\n### Protocol 3: Regression Prevention\n\nBefore any change:\n\n```bash\n# Baseline testing\ncurl -s http://localhost:3000/api/health  # API works\npnpm tsc --noEmit                          # TypeScript valid\npnpm test                                   # Tests pass\n\n# After change - all must still work\n```\n\n---\n\n## Common Commands\n\n### Daily Development\n\n```bash\n# Start coding session\nclaude \"Review CLAUDE.md, then help me with [task]\"\n\n# Quick check before commit\nclaude \"Run pre-commit workflow\"\n\n# Verify a specific file\nclaude \"Review [file] for bugs, especially [concern area]\"\n```\n\n### Pre-Launch\n\n```bash\n# Full audit\nclaude \"Run full-audit workflow on src/\"\n\n# Browser QA\nclaude \"Use fullstack-qa-orchestrator to test http://localhost:3000\"\n\n# Security focus\nclaude \"Use security-auditor to deep scan src/api/\"\n```\n\n### Debugging\n\n```bash\n# Console monitoring\nclaude \"Use console-monitor to watch http://localhost:3000\"\n\n# Specific bug\nclaude \"Run bug-fix workflow for: [bug description]\"\n```\n\n---\n\n## Files Generated\n\nAll reports go to `.claude/audits/` (gitignored):\n\n| File | Source Agent |\n|------|--------------|\n| AUDIT_CODE.md | code-auditor |\n| AUDIT_SECURITY.md | bug-auditor |\n| AUDIT_SECURITY_DEEP.md | security-auditor |\n| AUDIT_DOCS.md | doc-auditor |\n| AUDIT_INFRA.md | infra-auditor |\n| AUDIT_UI_UX.md | ui-auditor |\n| AUDIT_DB.md | db-auditor |\n| AUDIT_PERF.md | perf-auditor |\n| AUDIT_DEPS.md | dep-auditor |\n| AUDIT_SEO.md | seo-auditor |\n| API_TEST_REPORT.md | api-tester |\n| DEPLOY_CHECK.md | deploy-checker |\n| ENV_REPORT.md | env-validator |\n| FIXES.md | fix-planner |\n| TEST_REPORT.md | test-runner |\n| AUDIT_BROWSER_QA.md | browser-qa-agent |\n| QA_SESSION_LOG.md | fullstack-qa-orchestrator |\n\n---\n\n## Quick Reference\n\n| Task | Command |\n|------|---------|\n| New project setup | `curl -s https://undeadlist.com/agents/setup.sh \\| bash` |\n| Enable browser QA | `/chrome` in Claude Code |\n| Pre-commit check | `\"Run pre-commit workflow\"` |\n| Full audit | `\"Run full-audit workflow on src/\"` |\n| Pre-deploy check | `\"Run pre-deploy workflow\"` |\n| Browser test | `\"Use browser-qa-agent to test localhost:3000\"` |\n| Full QA loop | `\"Use fullstack-qa-orchestrator...\"` |\n| New feature | `\"Run new-feature workflow for: [feature]\"` |\n| Bug fix | `\"Run bug-fix workflow for: [bug]\"` |\n| Release prep | `\"Run release-prep workflow for v1.0.0\"` |\n| Verify production-ready | `\"Use architect-reviewer...\"` |\n\n---\n\n## When AI Goes Rogue\n\n### Red Flags - STOP Immediately\n\n- Agent makes changes without asking\n- Agent installs packages without approval\n- Agent modifies config files unexpectedly\n- Agent claims success without verification\n- Agent skips checkpoints\n- Agent touches files outside scope\n\n### Recovery Commands\n\n```bash\n# Revert last change\ngit checkout -- [file]\n\n# Revert all changes\ngit reset --hard HEAD\n\n# Check what changed\ngit diff\n```\n\n### Enforcement Phrases\n\n```\n\"STOP. You didn't ask permission.\"\n\"STOP. That's not in scope.\"\n\"STOP. Verify it works before claiming done.\"\n\"STOP. One change at a time.\"\n```\n\n---\n\n## Uninstall\n\n```bash\n# Option A: One-liner (recommended)\ncurl -s https://raw.githubusercontent.com/undeadlist/claude-code-agents/main/uninstall.sh | bash\n\n# Option B: Manual removal\nrm -rf .claude/agents/ .claude/audits/ workflows/\n```\n\nThis removes:\n- `.claude/agents/` - agent prompt definitions\n- `.claude/audits/` - generated reports (gitignored)\n- `workflows/` - predefined workflow chains\n\n\u003e **Note:** CLAUDE.md is left in place. Remove manually if desired.\n\n---\n\n## Contributing\n\nFound a better protocol? Agent that should exist?\n\nPR it: https://github.com/undeadlist/claude-code-agents\n\n---\n\n## License\n\nMIT. Use it, fork it, enhance it.\n\n---\n\n**Built by [UndeadList](https://undeadlist.com)**\n*The indie software flea market*\n\n---\n\n**Disclaimer:** Claude Code is a product of [Anthropic PBC](https://anthropic.com). This project is not affiliated with or endorsed by Anthropic.\n\n---\n\n*\"Built, but undiscovered.\"*\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fundeadlist%2Fclaude-code-agents","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fundeadlist%2Fclaude-code-agents","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fundeadlist%2Fclaude-code-agents/lists"}