/ Directory / Playground / claude-code-skills
● Community levnikolaevich ⚡ Instant

claude-code-skills

by levnikolaevich · levnikolaevich/claude-code-skills

A 7-plugin suite for Claude Code with 135+ skills and 3 bundled MCPs (hash-verified editing, code graph, remote SSH) — covers the full software lifecycle.

claude-code-skills (levnikolaevich) bundles plugins for Agile workflow, documentation, audits, bootstrap, optimization, community, and environment setup. Ships hex-line (hash-verified edits), hex-graph (code knowledge graph), and hex-ssh MCPs. MIT, fully open source.

Why use it

Key features

Live Demo

What it looks like in practice

claude-code.replay ▶ ready
0/0

Install

Pick your client

~/Library/Application Support/Claude/claude_desktop_config.json  · Windows: %APPDATA%\Claude\claude_desktop_config.json
{
  "mcpServers": {
    "claude-code": {
      "command": "npx",
      "args": [
        "-y",
        "claude-code-skills"
      ],
      "_inferred": true
    }
  }
}

Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.

~/.cursor/mcp.json · .cursor/mcp.json
{
  "mcpServers": {
    "claude-code": {
      "command": "npx",
      "args": [
        "-y",
        "claude-code-skills"
      ],
      "_inferred": true
    }
  }
}

Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.

VS Code → Cline → MCP Servers → Edit
{
  "mcpServers": {
    "claude-code": {
      "command": "npx",
      "args": [
        "-y",
        "claude-code-skills"
      ],
      "_inferred": true
    }
  }
}

Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".

~/.codeium/windsurf/mcp_config.json
{
  "mcpServers": {
    "claude-code": {
      "command": "npx",
      "args": [
        "-y",
        "claude-code-skills"
      ],
      "_inferred": true
    }
  }
}

Same shape as Claude Desktop. Restart Windsurf to pick up changes.

~/.continue/config.json
{
  "mcpServers": [
    {
      "name": "claude-code",
      "command": "npx",
      "args": [
        "-y",
        "claude-code-skills"
      ]
    }
  ]
}

Continue uses an array of server objects rather than a map.

~/.config/zed/settings.json
{
  "context_servers": {
    "claude-code": {
      "command": {
        "path": "npx",
        "args": [
          "-y",
          "claude-code-skills"
        ]
      }
    }
  }
}

Add to context_servers. Zed hot-reloads on save.

claude mcp add claude-code -- npx -y claude-code-skills

One-liner. Verify with claude mcp list. Remove with claude mcp remove.

Use Cases

Real-world ways to use claude-code-skills

Bootstrap a new project with Clean Architecture via claude-code-skills

👤 Solo devs, startup founders ⏱ ~30 min intermediate

When to use: You want a structured starting point instead of a blank directory.

Prerequisites
  • Claude Code installed — claude.ai/code
Flow
  1. Install the marketplace
    Add the claude-code-skills marketplace and install 'Project Bootstrap'.✓ Copied
    → Plugin installed and commands available
  2. Scaffold
    Use the bootstrap skill to scaffold a Python FastAPI project with Clean Architecture: domain / application / infrastructure / interfaces layers.✓ Copied
    → Directory tree with the standard layering + example module
  3. Set up dev env
    Use Setup Environment to add pre-commit, linters, test runner.✓ Copied
    → Working make test out of the box

Outcome: A day-1 project with the structure you'd refactor toward eventually.

Get multi-model code review with claude-code-skills

👤 Teams that want cross-checked AI reviews ⏱ ~20 min advanced

When to use: Before merging significant PRs, you want more than one model's opinion.

Prerequisites
  • Codex and Gemini API keys configured — Plugin docs show env var names
Flow
  1. Run the review skill
    Use the Agile Workflow plugin's review skill on PR #123. Run Codex and Gemini in parallel; fall back to Claude if either times out.✓ Copied
    → Three-column review: security, architecture, style
  2. Aggregate disagreements
    Where do the models disagree? Highlight those for human review.✓ Copied
    → Disagreement list — the most valuable section

Outcome: A review that surfaces issues a single model would miss.

Combine with: github

Auto-generate and maintain docs with the documentation pipeline

👤 Teams behind on docs ⏱ ~40 min intermediate

When to use: Your README and architecture docs haven't been updated in months.

Flow
  1. Run framework detection + initial gen
    Run the Documentation Pipeline on this repo. Detect framework, generate README, architecture.md, and contributing.md.✓ Copied
    → Draft docs reflecting actual code
  2. Set up refresh on CI
    Add a CI job that regenerates doc sections marked <!-- AUTOGEN --> on every merge.✓ Copied
    → Docs stay fresh

Outcome: Living docs backed by the same codebase source of truth.

Run a full codebase audit (security, quality, architecture)

👤 Tech leads inheriting a codebase ⏱ ~60 min intermediate

When to use: First week on a new repo — you need a map and a risk list.

Flow
  1. Audit
    Run the Codebase Audit Suite. Return security, quality, architecture, test-coverage findings ranked by impact.✓ Copied
    → Prioritized findings with file citations
  2. Triage into backlog
    Convert top-10 findings into GitHub issues with labels.✓ Copied
    → Backlog populated

Outcome: A clear picture of what to fix first.

Combine with: github

Combinations

Pair with other MCPs for X10 leverage

claude-code + github

Audit findings → issues → PRs via the same suite

Run the audit, create issues for the top 10 findings, then draft PRs for the first 3.✓ Copied
claude-code + hyper

Replace miscellaneous MCPs with a single hyper-mcp host

Keep hex-line/graph/ssh; retire unrelated npm MCPs in favor of signed plugins in hyper-mcp.✓ Copied

Tools

What this MCP exposes

ToolInputsWhen to callCost
hex-line (MCP) file edit ops with hash verification When correctness of edits matters — avoids 'phantom edit' bugs 0
hex-graph (MCP) codebase queries Navigate symbol references, callers, imports 0 after index
hex-ssh (MCP) host, command Run commands on dev/prod boxes via SSH with audit 0

Cost & Limits

What this costs to run

API quota
Uses your LLM quotas plus whatever the MCPs reach
Tokens per call
Audit skills can use 20k+ tokens — worth it once per project
Monetary
Free, MIT
Tip
Install only the plugins you need — all 7 active at once is overwhelming

Security

Permissions, secrets, blast radius

Credential storage: SSH keys for hex-ssh via standard ~/.ssh; no secrets stored by the suite
Data egress: Depends on enabled plugins
Never grant: broad SSH to prod without audit

Troubleshooting

Common errors and fixes

Plugin commands missing after install

Restart Claude Code; plugin registry is loaded at session start

Multi-model review times out on one model

Plugin falls back to Claude — check logs for which model failed and why

hex-ssh fails with auth error

SSH config must work without interactive prompts; use keys not passwords

Verify: ssh -o BatchMode=yes user@host true

Alternatives

claude-code-skills vs others

AlternativeWhen to use it insteadTradeoff
Individual MCPs composed yourselfYou prefer hand-picked tools over a suiteMore setup; no shared orchestration model
Cursor Rules + built-insYou use Cursor, not Claude CodeDifferent ecosystem

More

Resources

📖 Read the official README on GitHub

🐙 Browse open issues

🔍 Browse all 400+ MCP servers and Skills