nobrainer-claude-skills

A collection of Claude Code skills that I build and use daily. Publishing them here so others can benefit too.

This is a personal repo — I'm the sole maintainer. If you find a skill useful, feel free to grab it. If you have ideas or find bugs, open an issue.

Skills

nobrainer-fast-audit

Universal security diagnostic skill for Claude Code. Cross-platform: macOS, Linux, Windows, VPS.

Commands:

• /safety-audit — Full system security audit (OS posture, agent config, credentials, network, processes)
• /safety-check-skill <path> — Vet a skill/plugin before installation (risk score 0-100)
• /safety-scan — Quick IOC scan (C2 connections, malware artifacts, suspicious processes)

Covers:

• macOS: SIP, Gatekeeper, FileVault, Firewall, XProtect, Launch Agents
• Linux: SELinux/AppArmor, UFW/iptables, LUKS, unattended upgrades, systemd
• Windows: Defender, UAC, BitLocker, Windows Firewall, SmartScreen
• VPS: SSH hardening, fail2ban, rootkit detection, open ports
• AI Agent Platforms: Claude Code, OpenClaw, and other agent frameworks
• OWASP Agentic Top 10 (2026) hardening guide (ASI01-ASI10)
• Known threats: ClawHavoc campaign, MCP server CVEs, supply chain patterns

nobrainer-polymarket

Interact with Polymarket prediction markets via the official polymarket-cli. Browse markets, check prices, place orders, manage positions, and redeem winnings — all from the terminal.

Use when: searching markets, checking prices/order books, trading, viewing portfolios, redeeming won positions, or any Polymarket task via CLI.

Includes:

• Installation instructions (install script, manual download with checksum verification, build from source)
• Authentication setup (wallet import, config file, env vars, signature types)
• All workflows: market research, trading, order management, CTF redeem, leaderboard
• Full command reference in references/commands.md
• NegRisk vs standard market guidance (most binary markets are NegRisk)

nobrainer-team-builder

Dynamically assemble a team of expert subagents for any task from a catalog of 401 agents across 27 categories. Agents are spawned on-demand as temporary subagents — zero permanent context cost.

Trigger: "nbteam", "team builder", "build team", "assemble team"

How it works:

1. Reads your task/request
2. Selects 2-5 relevant categories → loads only those category JSONs
3. Picks best 3-10 agents from loaded categories
4. Reads full agent .md only for selected agents
5. Spawns each as a temporary subagent via the Agent tool
6. Collects results and synthesizes a unified answer

Categories include: AI specialists, API/GraphQL, blockchain/Web3, data/AI, database, DevOps, security, programming languages (49 agents), development tools, documentation, and 17 more.

Agent definitions sourced from davila7/claude-code-templates.

nobrainer-autopilot

Autonomous CI/CD workflow for any project. Converts TODOs to GitHub issues, spawns expert teams (min 5 agents via nobrainer-team-builder), implements, creates PRs with Copilot review gate, merges — all hands-free.

Trigger: "autopilot", "go go", "dzialaj sam", "lec sam", "cisnij"

How it works:

1. Discovery: Spawns 3 agents to review codebase, find TODOs, assess test coverage
2. Planning: Presents prioritized list, converts TODOs to GitHub issues after approval
3. Autonomous loop: For each issue — spawns 5-agent team, implements, creates PR, waits for Copilot review, replies to all comments, resolves threads, merges, deploys
4. Quality gates: DRY/KISS/SOLID/YAGNI, max 300 lines/file (350 acceptable), tests pass, Copilot reviewed with 0 unresolved threads

Prerequisites: nobrainer-team-builder installed, GitHub Copilot code review enabled on repo.

nobrainer-continuous-improvement

Upgrades any project's CLAUDE.md with proven workflow rules — Workflow Orchestration (6 rules), Task Management (6 steps), and Core Principles (3 rules). Merge-safe: only adds missing sections, never overwrites existing content.

Trigger: "continuous improvement", "upgrade CLAUDE.md", "dodaj zasady do CLAUDE.md", "dopisz workflow rules"

Adds:

• Workflow Orchestration: Plan Mode, Subagent Strategy, Self-Improvement Loop, Verification, Elegance, Autonomous Bug Fixing
• Task Management: plan → verify → track → explain → document → capture lessons
• Core Principles: Simplicity First, No Laziness, Minimal Impact
• Scaffolds tasks/todo.md + tasks/lessons.md

nobrainer-memory

Install persistent semantic memory for Claude Code using memsearch. Every session is auto-captured as markdown notes, with relevant context injected on every prompt via local Ollama embeddings — no API key needed.

Trigger: "install memory", "setup memsearch", "nobrainer-memory", "dodaj pamiec do claude"

What gets installed:

• memsearch Python CLI (PyPI)
• nomic-embed-text Ollama model — local embeddings
• memsearch ccplugin registered in Claude Code plugins
• Config: ~/.memsearch/config.toml

Supports global (~/.memsearch/memory/) or per-project memory scope.

nobrainer-starter

Project bootstrapper — creates AGENTS.md and CLAUDE.md with engineering standards and workflow rules in any project directory.

Commands:

• /nobrainer-starter — Bootstrap current project with all standards
• or: "setup project", "init project", "create AGENTS.md", "bootstrap project standards"

Both files receive identical content — works with Claude Code, Codex, Kimi, Cursor, or any AI coding assistant that reads project config files.

Creates:

• AGENTS.md + CLAUDE.md — identical content covering:
  • Engineering principles: DRY, KISS, SOLID, YAGNI, Clean Code, error handling, testability, communication ethics (no docs unless asked, no emojis, single method approach)
  • Workflow rules: Plan Mode, Subagent Strategy, Verification Before Done, Task Management, Git Rules
  • Safety Rules: destructive commands (rm -rf, DROP TABLE, git reset --hard, force push, wiping directories) require explicit confirmation before execution
• tasks/todo.md — task tracker stub
• tasks/lessons.md — lessons log stub

Merge-safe: if files already exist, only missing sections are added — existing content is never overwritten.

Installation

Copy any skill directory into ~/.claude/skills/:

# Clone the repo
git clone https://github.com/nobrainer-tech/nobrainer-claude-skills.git

# Install a skill
cp -r nobrainer-claude-skills/nobrainer-fast-audit ~/.claude/skills/
cp -r nobrainer-claude-skills/nobrainer-starter ~/.claude/skills/
cp -r nobrainer-claude-skills/nobrainer-polymarket ~/.claude/skills/
cp -r nobrainer-claude-skills/nobrainer-team-builder ~/.claude/skills/
cp -r nobrainer-claude-skills/nobrainer-memory ~/.claude/skills/
cp -r nobrainer-claude-skills/nobrainer-autopilot ~/.claude/skills/
cp -r nobrainer-claude-skills/nobrainer-continuous-improvement ~/.claude/skills/

# Or symlink (auto-updates with git pull)
ln -s "$(pwd)/nobrainer-claude-skills/nobrainer-fast-audit" ~/.claude/skills/nobrainer-fast-audit
ln -s "$(pwd)/nobrainer-claude-skills/nobrainer-starter" ~/.claude/skills/nobrainer-starter
ln -s "$(pwd)/nobrainer-claude-skills/nobrainer-polymarket" ~/.claude/skills/nobrainer-polymarket
ln -s "$(pwd)/nobrainer-claude-skills/nobrainer-team-builder" ~/.claude/skills/nobrainer-team-builder
ln -s "$(pwd)/nobrainer-claude-skills/nobrainer-memory" ~/.claude/skills/nobrainer-memory
ln -s "$(pwd)/nobrainer-claude-skills/nobrainer-autopilot" ~/.claude/skills/nobrainer-autopilot
ln -s "$(pwd)/nobrainer-claude-skills/nobrainer-continuous-improvement" ~/.claude/skills/nobrainer-continuous-improvement

Then restart Claude Code — the skill will be available immediately.

License

MIT