66 skills for Claude Code and Codex CLI covering code review, planning, UI/UX, testing, automation, SaaS & DevOps, and NTM swarm coordination. The same SKILL.md format works in both CLIs, and the installer writes to both by default.
Includes skills created or inspired by Jeffrey Emanuel alongside project workflow and swarm coordination skills.
claude-skills-md.zip ships every skill in one archive. Unzip it into either or both CLI skill directories:
curl -fsSL https://github.com/danzam98/claude-skills-toolkit/raw/master/claude-skills-md.zip -o claude-skills-md.zip
# Claude Code
unzip claude-skills-md.zip -d ~/.claude/skills/
# Codex CLI
unzip claude-skills-md.zip -d ~/.codex/skills/Mac/Linux:
git clone https://github.com/danzam98/claude-skills-toolkit.git
cd claude-skills-toolkit
./install.shWindows:
git clone https://github.com/danzam98/claude-skills-toolkit.git
cd claude-skills-toolkit
install.bat./install.sh # Install all skills to Claude + Codex (interactive)
./install.sh --all --force # Install all, overwrite existing
./install.sh fresh-eyes bug-hunt # Install specific skills (Claude + Codex)
./install.sh --target=claude fresh-eyes # Install only to Claude Code
./install.sh --target=codex bug-hunt # Install only to Codex CLI
./install.sh -f peer-review # Install with overwrite
./install.sh --list # List available skills
./install.sh --help # Show helpBy default, every install writes to both ~/.claude/skills/ and ~/.codex/skills/. Use --target=claude or --target=codex to install to only one CLI.
Mac/Linux:
# One-liner to install a specific skill into both CLIs (replace SKILL_NAME)
mkdir -p ~/.claude/skills/SKILL_NAME ~/.codex/skills/SKILL_NAME && \
curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/SKILL_NAME/SKILL.md -o ~/.claude/skills/SKILL_NAME/SKILL.md && \
cp ~/.claude/skills/SKILL_NAME/SKILL.md ~/.codex/skills/SKILL_NAME/SKILL.md
# Examples:
mkdir -p ~/.claude/skills/fresh-eyes ~/.codex/skills/fresh-eyes && \
curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/fresh-eyes/SKILL.md -o ~/.claude/skills/fresh-eyes/SKILL.md && \
cp ~/.claude/skills/fresh-eyes/SKILL.md ~/.codex/skills/fresh-eyes/SKILL.mdOr clone and copy specific skills:
git clone https://github.com/danzam98/claude-skills-toolkit.git
cp -r claude-skills-toolkit/skills/fresh-eyes ~/.claude/skills/
cp -r claude-skills-toolkit/skills/fresh-eyes ~/.codex/skills/- Claude Code and/or Codex CLI installed and configured. Codex CLI version 0.118.0 or newer reads
~/.codex/skills/using the same SKILL.md format as Claude Code.
NTM swarm skills require additional tooling for full functionality. All tools fall back gracefully when absent, but the full workflow needs:
| Tool | Purpose | Install |
|---|---|---|
| br (beads_rust) | Issue tracking via .beads/ files |
github.com/Dicklesworthstone/beads_rust |
| bv (beads_viewer) | Graph-aware task triage and prioritization | github.com/Dicklesworthstone/beads_viewer |
| MCP Agent Mail | Inter-agent messaging for swarm coordination | Configure as an MCP server in Claude Code settings |
| tmux | Terminal multiplexer for running multiple agents | brew install tmux / apt install tmux |
NTM swarm skills work fine for single-agent workflows without tmux — just invoke them directly in Claude Code. tmux is only needed for multi-agent parallel sessions.
| Skill | Command | Description |
|---|---|---|
| Fresh Eyes | /fresh-eyes |
Automated code review that catches bugs, security issues, and logic errors before they ship |
| Bug Hunt | /bug-hunt |
Deep codebase investigation to find hidden bugs, security holes, and performance issues |
| Bug Hunter | /bug-hunter |
Randomly explore and trace code execution flows to find obvious mistakes and issues |
| Peer Review | /peer-review |
Structured code review covering bugs, security, performance, style, and testing |
| Peer Code Reviewer | /peer-code-reviewer |
Cross-agent code review to catch issues from parallel work streams |
| Plan Review | /plan-review |
Validate implementation plans to catch errors and bad assumptions before coding |
Install just this category:
./install.sh -f fresh-eyes bug-hunt bug-hunter peer-review peer-code-reviewer plan-review| Skill | Command | Description |
|---|---|---|
| Idea Wizard | /idea-wizard |
Generate and evaluate improvement ideas with data-driven prioritization |
| 100-to-10 Filter | /hundred-to-ten-filter |
Generate 100 ideas, then ruthlessly filter to the 10 most brilliant |
| Next Best Move | /next-best-move |
Find the single smartest, most radically innovative addition to the project right now |
| PRD Generator | /prd |
Generate comprehensive Product Requirements Documents for features and projects |
| Premortem Planner | /premortem-planner |
Imagine failure 6 months out and revise the plan to prevent it |
| Project Opinion | /project-opinion-elicitor |
Get honest, critical assessment of the project from the agent's perspective |
| System Weaknesses | /system-weaknesses |
Identify the weakest parts of the system that need improvement |
Install just this category:
./install.sh -f idea-wizard hundred-to-ten-filter next-best-move prd premortem-planner project-opinion-elicitor system-weaknesses| Skill | Command | Description |
|---|---|---|
| Stripe-Level UI | /stripe-level-ui |
Build polished UI/UX components with strong attention to visual quality |
| UI Polish | /ui-polish |
Iterative refinement passes for apps that already work and need to look better |
| UX Audit | /ux-audit |
Systematic UX evaluation using Nielsen heuristics and accessibility checks |
| React Component Generator | /react-component-generator |
React 19 components with TypeScript, accessibility, and tests built in |
| Interactive Visualization Creator | /interactive-visualization-creator |
Interactive visualizations for Next.js using SVG, Canvas, Three.js, and Framer Motion |
| Admin Page for Next.js Sites | /admin-page-for-nextjs-sites |
Cohesive Next.js SaaS admin cockpits covering permissions, audit, analytics, and ops |
Install just this category:
./install.sh -f stripe-level-ui ui-polish ux-audit react-component-generator interactive-visualization-creator admin-page-for-nextjs-sites| Skill | Command | Description |
|---|---|---|
| Code Reorganizer | /code-reorganizer |
Restructure scattered code files into a sensible, intuitive folder structure |
| Deep Performance Audit | /deep-performance-audit |
Systematic identification of optimization opportunities with proof requirements |
| Stub Eliminator | /stub-eliminator |
Replace all stubs, placeholders, and mocks with production-ready code |
| De-Slopify | /de-slopify |
Remove telltale AI writing patterns from documentation and text |
| Multi-Model Synthesis | /multi-model-synthesis |
Blend competing LLM outputs into a superior hybrid plan |
Install just this category:
./install.sh -f code-reorganizer deep-performance-audit stub-eliminator de-slopify multi-model-synthesis| Skill | Command | Description |
|---|---|---|
| E2E Pipeline Validator | /e2e-pipeline-validator |
Prove the entire system works with real data, no mocks allowed |
| Deployment Verifier | /deployment-verifier |
Verify live deployment works with automated browser testing |
| Testing Fuzzing | /testing-fuzzing |
Coverage-guided, structure-aware, differential, stateful, and protocol fuzzing with AFL++ CMPLOG, persistent mode, and sanitizers |
| Testing Golden Artifacts | /testing-golden-artifacts |
Snapshot and approval test suites that freeze known-good outputs and catch regressions via exact comparison |
| Testing Real-Service E2E (No Mocks) | /testing-real-service-e2e-no-mocks |
Mock-free integration and E2E tests against real databases and APIs with transaction rollback isolation |
| E2E Testing for Webapps | /e2e-testing-for-webapps |
Next.js, Playwright, and Supabase: OAuth bypass via test users, interactive debugging, visual QA, Electron testing |
| Test E2E Webapps | /test-e2e-webapps |
Next.js, Playwright, and Supabase: page objects, console monitoring, visual regression, test user provisioning |
Install just this category:
./install.sh -f e2e-pipeline-validator deployment-verifier testing-fuzzing testing-golden-artifacts testing-real-service-e2e-no-mocks e2e-testing-for-webapps test-e2e-webapps| Skill | Command | Description |
|---|---|---|
| Agent Swarm Launcher | /agent-swarm-launcher |
Initialize multiple agents with full context and coordination protocols |
| Robot Mode Maker | /robot-mode-maker |
Create an agent-optimized CLI interface for any project |
| CLI Error Tolerance | /cli-error-tolerance |
Make CLI tools forgiving of minor syntax issues for agent ergonomics |
| Git Committer | /git-committer |
Intelligently commit all changed files in logical groupings with detailed messages |
| Ralph | /ralph |
Convert PRDs to JSON format for the Ralph autonomous agent system |
| Repeatedly Apply Skill | /repeatedly-apply-skill |
Iteratively apply any skill N times with progressive deepening and subagent delegation |
| Beads Workflow | /beads-workflow |
Convert markdown plans into actionable beads with dependencies, polishing, and swarm-ready task graphs |
Install just this category:
./install.sh -f agent-swarm-launcher robot-mode-maker cli-error-tolerance git-committer ralph repeatedly-apply-skill beads-workflow| Skill | Command | Description |
|---|---|---|
| Deep Project Primer | /deep-project-primer |
Essential first step to fully understand a project before any work |
| README Reviser | /readme-reviser |
Update documentation for recent changes, framing them as how it always was |
| Gemini Research | /gemini-grounded-research |
Web-backed research using Gemini with Google Search grounding for real-time, citation-backed answers |
| BD to BR Migration | /bd-to-br-migration |
Migrate docs from bd (beads) to br (beads_rust) |
| Cost Estimate | /cost-estimate |
Estimate development cost of a codebase with LOC analysis, market rates, org overhead, team multipliers, and Claude ROI |
Install just this category:
./install.sh -f deep-project-primer readme-reviser gemini-grounded-research bd-to-br-migration cost-estimateEnd-to-end coverage for shipping and operating a subscription SaaS — payments, billing hardening, analytics, growth, infrastructure, auth, and customer support. Most skills assume Next.js on the front end and Supabase on the back end, but the patterns generalize. Pairs naturally with the UI/UX category for product surfaces and Testing & Deployment for verification.
| Skill | Command | Description |
|---|---|---|
| Stripe Checkout | /stripe-checkout |
Stripe and PayPal subscription checkout for SaaS apps — recurring billing, webhook handling, Next.js with Rust CLI companions |
| SaaS Billing Patterns | /saas-billing-patterns-for-stripe-and-paypal |
Audit, harden, or implement SaaS subscription billing — webhooks, dunning, duplicate-charge prevention, MRR, security |
| SaaS Customer Analytics | /saas-customer-analytics |
MRR, churn, behavioral scoring, Monte Carlo revenue projections, and intervention design for subscription businesses |
| Supabase | /supabase |
Configure Supabase for Next.js SaaS — Drizzle + pooler, Google-only Auth, RLS, and cost optimization |
| GA4 | /ga4 |
Configure GA4 + GTM for Next.js SaaS — funnel tracking, scroll depth, CTA clicks, conversions, and feature usage |
| A/B Testing | /ab-testing |
Home-rolled A/B testing platform with server-side variant assignment, Bayesian analysis, Rust-WASM stats, and contextual bandits |
| SEO for SaaS | /seo-for-saas-businesses |
Multi-phase SEO + GEO program — technical SEO, Core Web Vitals, AI Overview citations, schema, hreflang, programmatic templates |
| SaaS CLI Auth Flow | /saas-cli-auth-flow |
CLI-to-web auth — PKCE OAuth, RFC 8628 device code, token lifecycle, and secure local storage |
| Support Ticketing System | /user-support-ticketing-system-for-saas |
Production-grade support ticketing — schema, SLA engine, APIs, queues, admin/user UI, email, cron |
| Support Triage | /user-support-triage-for-saas-and-open-source-projects |
Triage support across any project — DB tickets, GitHub issues, Zendesk, Intercom, SLA breaches, refunds, GDPR/CCPA |
Install just this category:
./install.sh -f stripe-checkout saas-billing-patterns-for-stripe-and-paypal saas-customer-analytics supabase ga4 ab-testing seo-for-saas-businesses saas-cli-auth-flow user-support-ticketing-system-for-saas user-support-triage-for-saas-and-open-source-projectsNTM (Named Tmux Manager) is a multi-agent orchestration pattern where multiple Claude agents work in parallel on a single project, coordinated through a shared task queue (beads), MCP Agent Mail for inter-agent messaging, and a small number of sticky coordination roles when necessary. These skills are purpose-built for NTM swarm workflows.
NTM skills are framework-agnostic — they work with any project type (Next.js, Rails, Vite, HTML mockups, CLIs, etc.). Project-specific commands and paths live in
AGENTS.md; the skills contain workflow, recovery, and coordination logic. They prefer the project’s own authority chain and machine-readable surfaces, whether that is./robotor something else.
| Skill | Command | Description |
|---|---|---|
| NTM Project Prep | /ntm-project-prep |
Read AGENTS.md, follow the project’s authority chain, inspect live project surfaces, and recover the correct identity lease before work starts |
| NTM Start Agent | /ntm-start-agent |
Resume a fungible worker loop that checks mail, claims work, reserves files, reviews, and keeps moving without waiting for re-prompts |
| NTM Next Bead | /ntm-next-bead |
Re-enter the worker loop by checking coordination state, continuing owned work first, or taking the next actionable task |
| NTM Git Manager | /ntm-git-manager |
Run the sticky git-coordination role for a swarm, preserve the same identity lease, and keep landing work clean |
| NTM Commit All | /ntm-commit-all |
Commit all pending changes in logical groupings with beads sync; notifies git manager to push |
| NTM Reread Agents | /ntm-reread-agents |
Recover cleanly after compaction by reclaiming the same identity lease, re-reading authority surfaces, and reconstructing current work before resuming |
| NTM Review Own | /ntm-review-own |
Agent self-review quality gate using project check commands from AGENTS.md |
| NTM Review Others | /ntm-review-others |
Perform cross-agent review and fresh-eyes exploration to keep the swarm productive when direct implementation work is temporarily thin |
| NTM Bug Hunt | /ntm-bug-hunt |
Random codebase exploration using ./robot files for file discovery, then deep bug investigation |
| NTM Test Coverage | /ntm-test-coverage |
Audit and expand test coverage with comprehensive logging |
| NTM UI Polish | /ntm-ui-polish |
UI/UX refinement pass for world-class visual polish |
| NTM Unstall | /ntm-unstall |
Recover a stalled swarm queue by conservatively reopening abandoned work and returning fungible workers to productive motion |
Install just this category:
./install.sh -f ntm-project-prep ntm-start-agent ntm-next-bead ntm-git-manager ntm-commit-all ntm-reread-agents ntm-review-own ntm-review-others ntm-bug-hunt ntm-test-coverage ntm-ui-polish ntm-unstall| Skill | Command | Description |
|---|---|---|
| RCH | /rch |
Remote compilation helper for offloading cargo/gcc/bun builds to workers |
Install just this category:
./install.sh -f rchAutomatically review your code changes to catch bugs before they ship.
When to use:
- After writing or modifying code
- Before committing changes to git
- When you want an extra layer of quality assurance
What it catches:
- Logic errors and off-by-one bugs
- Missing null checks and error handling
- Security vulnerabilities (SQL injection, XSS, auth bypass)
- Performance issues (N+1 queries, unnecessary loops)
/fresh-eyes
Proactively explore your codebase to find hidden bugs, security holes, and performance issues.
When to use:
- Before major releases
- When investigating reliability issues
- For security audits
/bug-hunt # Explore entire codebase
/bug-hunt app/auth # Focus on specific area
Generate and evaluate improvement ideas for your project with data-driven prioritization.
How it works:
- Generates 30 diverse ideas across categories
- Evaluates each on effort vs. impact
- Winnows to top 5 with detailed rationale
- Offers to implement the best idea
/idea-wizard # General improvements
/idea-wizard "checkout flow" # Specific feature
Reads your AGENTS.md and project plan, then uses deep extended thinking to identify the one most impactful addition you could make right now.
What makes it different from Idea Wizard:
- Produces exactly one answer, not a ranked list
- Forces a strategic lens: what unlocks the most, not just what's useful
- Requires grounding in the actual project state first
/next-best-move
Systematic identification of optimization opportunities with proof requirements.
Methodology:
- Baseline metrics first (p50/p95/p99 latency, throughput)
- Profile before proposing changes
- Isomorphism proof per change
- One optimization per change for clear attribution
/deep-performance-audit
Research using Gemini with Google Search grounding for real-time, citation-backed answers.
Requirements: Set GEMINI_API_KEY environment variable
# Get free API key: https://aistudio.google.com/apikey
export GEMINI_API_KEY="your-key-here"/gemini-grounded-research "What are the latest React 19 features?"
Analyze any codebase and produce a professional development cost estimate suitable for stakeholders, investors, or clients.
What it calculates:
- Lines of code by language/complexity category (Swift, C++, Metal, tests, etc.)
- Base development hours using industry productivity benchmarks
- Overhead multipliers (architecture, debugging, code review, docs, testing)
- Realistic calendar time across company types (solo to enterprise)
- Market-rate research via live WebSearch for current 2025 rates
- Full team cost with supporting roles (PM, design, QA, DevOps, etc.)
- Claude ROI analysis: value per Claude active hour vs. human developer cost
Output includes:
- Low/average/high scenario cost tables
- Calendar time with organizational overhead (standups, meetings, context-switching)
- Role breakdown for Growth Company and Enterprise scenarios
- Speed multiplier and net savings vs. hiring a human developer
/cost-estimate
Orchestrate N passes of any skill against a target, each delegated to a fresh subagent with a unique mission.
How it works:
- Reads the target skill and generates N domain-specific missions
- Creates a
.skill-loop-progress.mdtracking file - Delegates each pass to a fresh subagent (no accumulated fatigue)
- Verifies, commits, and logs each pass before proceeding
- Stops on convergence, thrashing, or pass cap
When to use:
- When one pass of a skill isn't enough
- For iterative polish (UI, docs, bug scanning)
- When you want systematic, progressive deepening
/repeatedly-apply-skill 10 /ui-polish # 10 passes of UI polish
/repeatedly-apply-skill 5 /ubs # 5 passes of bug scanning
Convert markdown plans into actionable beads (tasks) with dependencies using the br CLI.
How it works:
- Takes a markdown plan and converts it to granular beads
- Adds dependency structure (what blocks what)
- Iterative polishing (6-9 rounds) until steady-state
- Validates with
br dep cyclesandbv --robot-insights
When to use:
- Before starting implementation of a plan
- When coordinating multi-agent swarms
- To bridge planning to execution
Includes reference files:
BEAD-ANATOMY.md— What makes a good beadPROMPTS.md— Complete prompt reference for conversion and polishingTROUBLESHOOTING.md— Worktree errors, sync issues, migration guide
/beads-workflow
NTM skills separate workflow logic from project specifics:
- Skills encode: process steps, reasoning patterns, decision logic, coordination protocols
AGENTS.mdencodes: file paths, package manager, check commands, tech stack, design tokens
This means the same skill works identically on a Next.js app, a Vite HTML prototype, a Rails API, or a CLI tool — the skill never assumes a framework. When an agent reads AGENTS.md at startup, it learns everything project-specific; the skills just tell it how to behave.
Project-local machine-readable surfaces: NTM skills prefer whatever live surfaces the project declares in AGENTS.md, such as ./robot or other CLI/status tools. Use /robot-mode-maker when you want a project-local CLI, but the NTM suite is not hardcoded to require it.
Fungible workers, sticky exceptions: Workers are generalists that pull from the same actionable queue. Sticky roles such as the git manager are deliberate exceptions and should preserve the same identity across compaction or long idle periods.
Git discipline: Worker agents never push to remote. The dedicated git manager (initialized with /ntm-git-manager) is the sole agent that runs git push when the project’s policy assigns pushes to that role. Worker agents commit locally and notify the git manager via MCP Agent Mail.
1. /deep-project-primer # Understand codebase
2. /idea-wizard # Explore options
3. /prd # Document requirements
4. /plan-review # Validate approach
5. [Implement feature]
6. /fresh-eyes # Quick check
7. /peer-review # Comprehensive review
8. /git-committer # Commit logically
1. /bug-hunt # Find hidden issues
2. /stub-eliminator # Ensure completeness
3. /e2e-pipeline-validator # Prove it works
4. /deployment-verifier # Test live deployment
1. /agent-swarm-launcher # Initialize agents
2. [Agents work in parallel]
3. /peer-code-reviewer # Cross-review work
4. /git-committer # Commit all changes
1. /ntm-project-prep # Read AGENTS and recover the correct identity
2. /ntm-git-manager # Keep one sticky landing and push role
3. /ntm-start-agent # Start each fungible worker loop
4. /ntm-next-bead # Re-enter task pickup after a completed bead
5. /ntm-review-own # Self-review quality gate before closing each bead
6. /ntm-review-others # Cross-review or fresh-eyes fallback when the queue is thin
7. /ntm-reread-agents # Recover cleanly after compaction
8. /ntm-unstall # Reopen only confirmed abandoned work when the queue stalls
After cloning, the simplest install path is ./install.sh -f <skill-names>, which writes to both ~/.claude/skills/ and ~/.codex/skills/ by default. Add --target=claude or --target=codex to limit one CLI. The curl commands below only target ~/.claude/skills/ and only work for skills whose only file is SKILL.md; for Codex installs or for skills with references/, scripts/, templates/, or examples/ subdirectories, use ./install.sh instead.
# Code Quality
mkdir -p ~/.claude/skills/fresh-eyes && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/fresh-eyes/SKILL.md -o ~/.claude/skills/fresh-eyes/SKILL.md
mkdir -p ~/.claude/skills/bug-hunt && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/bug-hunt/SKILL.md -o ~/.claude/skills/bug-hunt/SKILL.md
mkdir -p ~/.claude/skills/bug-hunter && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/bug-hunter/SKILL.md -o ~/.claude/skills/bug-hunter/SKILL.md
mkdir -p ~/.claude/skills/peer-review && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/peer-review/SKILL.md -o ~/.claude/skills/peer-review/SKILL.md
mkdir -p ~/.claude/skills/peer-code-reviewer && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/peer-code-reviewer/SKILL.md -o ~/.claude/skills/peer-code-reviewer/SKILL.md
mkdir -p ~/.claude/skills/plan-review && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/plan-review/SKILL.md -o ~/.claude/skills/plan-review/SKILL.md
# Ideation & Planning
mkdir -p ~/.claude/skills/idea-wizard && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/idea-wizard/SKILL.md -o ~/.claude/skills/idea-wizard/SKILL.md
mkdir -p ~/.claude/skills/hundred-to-ten-filter && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/hundred-to-ten-filter/SKILL.md -o ~/.claude/skills/hundred-to-ten-filter/SKILL.md
mkdir -p ~/.claude/skills/next-best-move && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/next-best-move/SKILL.md -o ~/.claude/skills/next-best-move/SKILL.md
mkdir -p ~/.claude/skills/prd && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/prd/SKILL.md -o ~/.claude/skills/prd/SKILL.md
mkdir -p ~/.claude/skills/premortem-planner && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/premortem-planner/SKILL.md -o ~/.claude/skills/premortem-planner/SKILL.md
mkdir -p ~/.claude/skills/project-opinion-elicitor && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/project-opinion-elicitor/SKILL.md -o ~/.claude/skills/project-opinion-elicitor/SKILL.md
mkdir -p ~/.claude/skills/system-weaknesses && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/system-weaknesses/SKILL.md -o ~/.claude/skills/system-weaknesses/SKILL.md
# UI/UX (multi-file skills, install.sh required)
./install.sh -f stripe-level-ui ui-polish ux-audit react-component-generator interactive-visualization-creator admin-page-for-nextjs-sites
# Refactoring & Optimization
mkdir -p ~/.claude/skills/code-reorganizer && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/code-reorganizer/SKILL.md -o ~/.claude/skills/code-reorganizer/SKILL.md
mkdir -p ~/.claude/skills/deep-performance-audit && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/deep-performance-audit/SKILL.md -o ~/.claude/skills/deep-performance-audit/SKILL.md
mkdir -p ~/.claude/skills/stub-eliminator && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/stub-eliminator/SKILL.md -o ~/.claude/skills/stub-eliminator/SKILL.md
mkdir -p ~/.claude/skills/de-slopify && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/de-slopify/SKILL.md -o ~/.claude/skills/de-slopify/SKILL.md
mkdir -p ~/.claude/skills/multi-model-synthesis && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/multi-model-synthesis/SKILL.md -o ~/.claude/skills/multi-model-synthesis/SKILL.md
# Testing & Deployment (single-file skills)
mkdir -p ~/.claude/skills/e2e-pipeline-validator && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/e2e-pipeline-validator/SKILL.md -o ~/.claude/skills/e2e-pipeline-validator/SKILL.md
mkdir -p ~/.claude/skills/deployment-verifier && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/deployment-verifier/SKILL.md -o ~/.claude/skills/deployment-verifier/SKILL.md
# Testing & Deployment (multi-file skills, install.sh required)
./install.sh -f testing-fuzzing testing-golden-artifacts testing-real-service-e2e-no-mocks e2e-testing-for-webapps test-e2e-webapps
# Automation & Agents
mkdir -p ~/.claude/skills/agent-swarm-launcher && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/agent-swarm-launcher/SKILL.md -o ~/.claude/skills/agent-swarm-launcher/SKILL.md
mkdir -p ~/.claude/skills/robot-mode-maker && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/robot-mode-maker/SKILL.md -o ~/.claude/skills/robot-mode-maker/SKILL.md
mkdir -p ~/.claude/skills/cli-error-tolerance && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/cli-error-tolerance/SKILL.md -o ~/.claude/skills/cli-error-tolerance/SKILL.md
mkdir -p ~/.claude/skills/git-committer && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/git-committer/SKILL.md -o ~/.claude/skills/git-committer/SKILL.md
mkdir -p ~/.claude/skills/ralph && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ralph/SKILL.md -o ~/.claude/skills/ralph/SKILL.md
mkdir -p ~/.claude/skills/repeatedly-apply-skill && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/repeatedly-apply-skill/SKILL.md -o ~/.claude/skills/repeatedly-apply-skill/SKILL.md
mkdir -p ~/.claude/skills/beads-workflow/references && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/beads-workflow/SKILL.md -o ~/.claude/skills/beads-workflow/SKILL.md && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/beads-workflow/references/PROMPTS.md -o ~/.claude/skills/beads-workflow/references/PROMPTS.md && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/beads-workflow/references/BEAD-ANATOMY.md -o ~/.claude/skills/beads-workflow/references/BEAD-ANATOMY.md && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/beads-workflow/references/TROUBLESHOOTING.md -o ~/.claude/skills/beads-workflow/references/TROUBLESHOOTING.md
# Workflow & Documentation
mkdir -p ~/.claude/skills/deep-project-primer && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/deep-project-primer/SKILL.md -o ~/.claude/skills/deep-project-primer/SKILL.md
mkdir -p ~/.claude/skills/readme-reviser && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/readme-reviser/SKILL.md -o ~/.claude/skills/readme-reviser/SKILL.md
mkdir -p ~/.claude/skills/gemini-grounded-research && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/gemini-grounded-research/SKILL.md -o ~/.claude/skills/gemini-grounded-research/SKILL.md
mkdir -p ~/.claude/skills/bd-to-br-migration && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/bd-to-br-migration/SKILL.md -o ~/.claude/skills/bd-to-br-migration/SKILL.md
mkdir -p ~/.claude/skills/cost-estimate && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/cost-estimate/SKILL.md -o ~/.claude/skills/cost-estimate/SKILL.md
# Infrastructure
mkdir -p ~/.claude/skills/rch && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/rch/SKILL.md -o ~/.claude/skills/rch/SKILL.md
# NTM Swarm Skills
mkdir -p ~/.claude/skills/ntm-project-prep && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-project-prep/SKILL.md -o ~/.claude/skills/ntm-project-prep/SKILL.md
mkdir -p ~/.claude/skills/ntm-start-agent && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-start-agent/SKILL.md -o ~/.claude/skills/ntm-start-agent/SKILL.md
mkdir -p ~/.claude/skills/ntm-next-bead && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-next-bead/SKILL.md -o ~/.claude/skills/ntm-next-bead/SKILL.md
mkdir -p ~/.claude/skills/ntm-git-manager && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-git-manager/SKILL.md -o ~/.claude/skills/ntm-git-manager/SKILL.md
mkdir -p ~/.claude/skills/ntm-commit-all && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-commit-all/SKILL.md -o ~/.claude/skills/ntm-commit-all/SKILL.md
mkdir -p ~/.claude/skills/ntm-reread-agents && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-reread-agents/SKILL.md -o ~/.claude/skills/ntm-reread-agents/SKILL.md
mkdir -p ~/.claude/skills/ntm-review-own && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-review-own/SKILL.md -o ~/.claude/skills/ntm-review-own/SKILL.md
mkdir -p ~/.claude/skills/ntm-review-others && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-review-others/SKILL.md -o ~/.claude/skills/ntm-review-others/SKILL.md
mkdir -p ~/.claude/skills/ntm-bug-hunt && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-bug-hunt/SKILL.md -o ~/.claude/skills/ntm-bug-hunt/SKILL.md
mkdir -p ~/.claude/skills/ntm-test-coverage && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-test-coverage/SKILL.md -o ~/.claude/skills/ntm-test-coverage/SKILL.md
mkdir -p ~/.claude/skills/ntm-ui-polish && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-ui-polish/SKILL.md -o ~/.claude/skills/ntm-ui-polish/SKILL.md
mkdir -p ~/.claude/skills/ntm-unstall && curl -fsSL https://raw.githubusercontent.com/danzam98/claude-skills-toolkit/master/skills/ntm-unstall/SKILL.md -o ~/.claude/skills/ntm-unstall/SKILL.mdProject templates live in the companion danzam98/toolkit repo, which provides:
| Resource | Location | Purpose |
|---|---|---|
AGENTS.md template |
toolkit/templates/AGENTS.md |
Universal agent rules template — fill in [TODO] sections for your project |
nextjs-static scaffold |
toolkit/templates/nextjs-static |
Next.js 16 + Tailwind v4 + Cloudflare Pages starter |
| Best-practices guides | toolkit/best-practices/ |
Authoritative guides agents reference during implementation |
See Setting Up a New Project for the full workflow.
Skills not showing up?
- Verify installation:
ls ~/.claude/skills/(Claude Code) orls ~/.codex/skills/(Codex CLI) - Restart Claude Code or Codex CLI
- Check file permissions
- For Codex CLI, confirm you are on version 0.118.0 or newer (
codex --version)
Gemini Grounded Research not working?
- Verify API key:
echo $GEMINI_API_KEY - Get key at: https://aistudio.google.com/apikey
Found a bug? Have an improvement? Edit the skill.md files and submit a PR.
MIT License - see LICENSE file for details.