From 330f307a1ab224d01766d7507a14ad6c387a9242 Mon Sep 17 00:00:00 2001 From: GAP Promoter Date: Thu, 28 May 2026 11:46:21 +0000 Subject: [PATCH 1/2] Add undeadlist/claude-code-agents to the registry --- .../undeadlist__claude-code-agents/README.md | 61 +++++++++++++++++++ .../metadata.json | 15 +++++ 2 files changed, 76 insertions(+) create mode 100644 agents/undeadlist__claude-code-agents/README.md create mode 100644 agents/undeadlist__claude-code-agents/metadata.json diff --git a/agents/undeadlist__claude-code-agents/README.md b/agents/undeadlist__claude-code-agents/README.md new file mode 100644 index 0000000..9e6a9fe --- /dev/null +++ b/agents/undeadlist__claude-code-agents/README.md @@ -0,0 +1,61 @@ +# claude-code-agents + +A complete end-to-end development workflow for solo developers building with **Next.js / React / TypeScript** using Claude Code. Built by [Paul @ UndeadList](https://undeadlist.com) from real-world production experience. + +## What It Does + +Gives a solo dev a full virtual engineering team, orchestrated through Claude Code's `Task()` API: + +- **24 specialized subagents** — each with a non-overlapping domain, designed to run in parallel +- **6 reusable workflow skills** — installable as slash commands (`/full-audit`, `/pre-commit`, `/pre-deploy`, `/new-feature`, `/bug-fix`, `/release-prep`) +- **Strict human-approval protocols** — AI agents cannot make changes without explicit checkpoint approval + +## Agent Categories + +| Category | Count | Purpose | +|---|---|---| +| Audit agents | 11 | Code, security, db, perf, SEO, deps, infra, UI, docs, bugs, APIs — run in parallel | +| Fix/implement agents | 4 | fix-planner, code-fixer, test-runner, test-writer | +| Browser QA agents | 4 | Chrome integration, UI testing, console monitoring, visual diff | +| Deploy agents | 2 | Pre-deploy validation, env config | +| Utility agents | 2 | PR writer, seed/test-data generator | +| Supervisor | 1 | architect-reviewer — final gate, oversees full pipeline | + +## Key Workflows + +```bash +# Full parallel audit — all 11 auditors simultaneously +claude "Run full-audit workflow on src/" + +# Pre-commit check +claude "Run pre-commit workflow" + +# TDD new feature +claude "Run new-feature workflow for: user authentication" + +# Pre-deployment validation +claude "Run pre-deploy workflow" +``` + +## Install as Plugin + +```bash +# Inside Claude Code +/plugin marketplace add undeadlist/claude-code-agents +/plugin install claude-code-agents@undeadlist-claude-code-agents +``` + +## Design Philosophy + +> *"You don't have a QA team. You don't have a code reviewer. You don't have time to waste on AI going rogue. This workflow is all of that — in a box, with guardrails."* + +All agents follow strict protocols: one change at a time, explicit scope, human checkpoint before destructive actions, regression testing after every fix. Audit reports write to `.claude/audits/` (gitignored). + +## Stack + +Optimized for Next.js / React / TypeScript / Prisma / Vercel. Stack-agnostic agents (security, code quality, docs, PR writer) work on any project. + +## Links + +- Repository: https://github.com/undeadlist/claude-code-agents +- Author: https://undeadlist.com diff --git a/agents/undeadlist__claude-code-agents/metadata.json b/agents/undeadlist__claude-code-agents/metadata.json new file mode 100644 index 0000000..b2c7df4 --- /dev/null +++ b/agents/undeadlist__claude-code-agents/metadata.json @@ -0,0 +1,15 @@ +{ + "name": "claude-code-agents", + "author": "undeadlist", + "description": "24 parallel audit/fix/QA/deploy agents + 6 workflow skills for Claude Code. Complete E2E dev workflow for solo devs building with Next.js and TypeScript.", + "repository": "https://github.com/undeadlist/claude-code-agents", + "path": "", + "version": "3.1.0", + "category": "developer-tools", + "tags": ["claude-code", "audit", "security", "code-quality", "testing", "browser-qa", "devops", "solo-dev", "typescript", "nextjs"], + "license": "MIT", + "model": "claude-sonnet-4-5-20250929", + "adapters": ["claude-code", "system-prompt"], + "icon": false, + "banner": false +} From 52675583d8c1a16574e1504a98a75a2e0eabc532 Mon Sep 17 00:00:00 2001 From: GAP Promoter Date: Sun, 31 May 2026 15:05:17 +0000 Subject: [PATCH 2/2] Update undeadlist/claude-code-agents: improve description, README, and point to fork with GAP files --- .../undeadlist__claude-code-agents/README.md | 117 ++++++++++++------ .../metadata.json | 6 +- 2 files changed, 83 insertions(+), 40 deletions(-) diff --git a/agents/undeadlist__claude-code-agents/README.md b/agents/undeadlist__claude-code-agents/README.md index 9e6a9fe..bd638a6 100644 --- a/agents/undeadlist__claude-code-agents/README.md +++ b/agents/undeadlist__claude-code-agents/README.md @@ -1,61 +1,104 @@ # claude-code-agents -A complete end-to-end development workflow for solo developers building with **Next.js / React / TypeScript** using Claude Code. Built by [Paul @ UndeadList](https://undeadlist.com) from real-world production experience. +**A complete E2E development workflow for solo-dev startups using Claude Code.** -## What It Does +Built by [Paul @ UndeadList](https://undeadlist.com), this plugin gives a solo developer a +full virtual engineering team — code reviewers, QA engineers, security auditors, and a +DevOps engineer — all orchestrated through Claude Code's `Task()` API with strict +human-approval protocols to keep AI in scope at every step. -Gives a solo dev a full virtual engineering team, orchestrated through Claude Code's `Task()` API: +## What It Does -- **24 specialized subagents** — each with a non-overlapping domain, designed to run in parallel -- **6 reusable workflow skills** — installable as slash commands (`/full-audit`, `/pre-commit`, `/pre-deploy`, `/new-feature`, `/bug-fix`, `/release-prep`) -- **Strict human-approval protocols** — AI agents cannot make changes without explicit checkpoint approval +Provides **24 specialized Claude Code subagents** and **6 workflow skills** that cover +the entire development lifecycle for Next.js / React / TypeScript projects (Prisma, npm/pnpm, Vercel). + +### Audit Agents (11 — run in parallel) +| Agent | Scope | +|---|---| +| `code-auditor` | Code quality, DRY, complexity | +| `bug-auditor` | Runtime bugs (not security) | +| `security-auditor` | OWASP deep scan (single authority) | +| `doc-auditor` | Documentation gap analysis | +| `infra-auditor` | Config, env vars, HTTP headers | +| `ui-auditor` | Accessibility, UX issues | +| `db-auditor` | N+1 queries, indexes, ORM misuse | +| `perf-auditor` | Bundle size, render performance | +| `dep-auditor` | Dependency vulnerabilities | +| `seo-auditor` | Meta tags, OpenGraph, structured data | +| `api-tester` | Endpoint validation, response contracts | + +### Fix / Implement Agents (4) +- **`fix-planner`** — prioritises audit findings into a `FIXES.md` action list +- **`code-fixer`** — implements approved fixes one at a time +- **`test-runner`** — validates fixes without regressions +- **`test-writer`** — auto-generates tests for new features (TDD) + +### Browser QA Agents (4) +- **`browser-qa-agent`** — navigates UI via Chrome, finds console errors +- **`fullstack-qa-orchestrator`** — end-to-end find → fix → verify loop +- **`console-monitor`** — real-time console error watching +- **`visual-diff`** — before/after screenshot comparison + +### Deploy Agents (2) +- **`deploy-checker`** — pre-deployment build and config validation +- **`env-validator`** — environment variable completeness + +### Utility Agents (2) +- **`pr-writer`** — generates detailed PR descriptions +- **`seed-generator`** — creates realistic test data + +### Supervisor (1) +- **`architect-reviewer`** — final gate; oversees audit → fix → re-audit loop + +## Workflow Skills (Slash Commands) + +| Skill | What It Does | +|---|---| +| `/full-audit` | All 11 auditors in parallel → fix-planner creates `FIXES.md` | +| `/pre-commit` | code-auditor + test-runner before committing | +| `/pre-deploy` | deploy-checker + env-validator + dep-auditor | +| `/new-feature` | test-writer → code-fixer → test-runner → browser-qa (TDD) | +| `/bug-fix` | Write failing test → fix → verify (regression prevention) | +| `/release-prep` | Full audit → fixes → deploy validation → pr-writer | + +## Install -## Agent Categories +```bash +# Inside Claude Code — install as a plugin (all 24 agents + 6 skills): +/plugin marketplace add undeadlist/claude-code-agents +/plugin install claude-code-agents@undeadlist-claude-code-agents +``` -| Category | Count | Purpose | -|---|---|---| -| Audit agents | 11 | Code, security, db, perf, SEO, deps, infra, UI, docs, bugs, APIs — run in parallel | -| Fix/implement agents | 4 | fix-planner, code-fixer, test-runner, test-writer | -| Browser QA agents | 4 | Chrome integration, UI testing, console monitoring, visual diff | -| Deploy agents | 2 | Pre-deploy validation, env config | -| Utility agents | 2 | PR writer, seed/test-data generator | -| Supervisor | 1 | architect-reviewer — final gate, oversees full pipeline | +Or from the terminal: +```bash +claude plugin install claude-code-agents@undeadlist-claude-code-agents --scope user +``` -## Key Workflows +## Example Usage ```bash -# Full parallel audit — all 11 auditors simultaneously +# Full parallel audit claude "Run full-audit workflow on src/" # Pre-commit check claude "Run pre-commit workflow" # TDD new feature -claude "Run new-feature workflow for: user authentication" +claude "Run new-feature workflow for: user settings page" -# Pre-deployment validation -claude "Run pre-deploy workflow" -``` - -## Install as Plugin +# Fix a specific bug +claude "Run bug-fix workflow for: checkout form submits twice on double-click" -```bash -# Inside Claude Code -/plugin marketplace add undeadlist/claude-code-agents -/plugin install claude-code-agents@undeadlist-claude-code-agents +# Full release prep +claude "Run release-prep workflow for v1.2.0" ``` -## Design Philosophy - -> *"You don't have a QA team. You don't have a code reviewer. You don't have time to waste on AI going rogue. This workflow is all of that — in a box, with guardrails."* - -All agents follow strict protocols: one change at a time, explicit scope, human checkpoint before destructive actions, regression testing after every fix. Audit reports write to `.claude/audits/` (gitignored). - ## Stack -Optimized for Next.js / React / TypeScript / Prisma / Vercel. Stack-agnostic agents (security, code quality, docs, PR writer) work on any project. +Optimised for **Next.js / React / TypeScript** with Prisma, npm/pnpm, and Vercel. +Stack-agnostic agents (`security-auditor`, `code-auditor`, `doc-auditor`, `pr-writer`, +`fix-planner`, `architect-reviewer`) work on any project. -## Links +## License -- Repository: https://github.com/undeadlist/claude-code-agents -- Author: https://undeadlist.com +MIT — use it, fork it, adapt it. diff --git a/agents/undeadlist__claude-code-agents/metadata.json b/agents/undeadlist__claude-code-agents/metadata.json index b2c7df4..2647bc3 100644 --- a/agents/undeadlist__claude-code-agents/metadata.json +++ b/agents/undeadlist__claude-code-agents/metadata.json @@ -1,12 +1,12 @@ { "name": "claude-code-agents", "author": "undeadlist", - "description": "24 parallel audit/fix/QA/deploy agents + 6 workflow skills for Claude Code. Complete E2E dev workflow for solo devs building with Next.js and TypeScript.", - "repository": "https://github.com/undeadlist/claude-code-agents", + "description": "24 Claude Code subagents + 6 workflow skills for solo-dev startups: parallel code audits, security scanning, browser QA, automated fix cycles, and release prep.", + "repository": "https://github.com/computer-agent/claude-code-agents", "path": "", "version": "3.1.0", "category": "developer-tools", - "tags": ["claude-code", "audit", "security", "code-quality", "testing", "browser-qa", "devops", "solo-dev", "typescript", "nextjs"], + "tags": ["claude-code", "subagents", "code-audit", "qa", "security", "solo-dev", "nextjs", "developer-workflow", "testing", "devops"], "license": "MIT", "model": "claude-sonnet-4-5-20250929", "adapters": ["claude-code", "system-prompt"],