Claude Code Skills & Plugins
A library of 362 production-ready skills, plugins, and agent skills for 13 AI coding tools, covering engineering, marketing, product, compliance, C-level advisory, and more.
Evidence shows the repo includes a security audit tool (skill-security-auditor) and SECURITY.md, but lacks explicit least-privilege design or user confirmation mechanisms. Data flow transparency is limited; no detailed explanation of how skills collect or process data. Sensitive data handling is addressed with secure coding practices and tests for hardcoded credentials, but not comprehensive. Dependency security is partially addressed with safety audit in CI, but no dependency manifest. External effects are not clearly documented; scripts may make network calls. Rollback is mentioned via semantic versioning and backward compatibility, but no concrete rollback mechanism. Source attribution is present for some third-party skills, but not detailed.
Self-consistency is moderate; README states 362 skills but description says 345, though structure is clear. Dependency availability is claimed (stdlib-only) but not verified. Failure messages are not well-documented; CI has error handling but no user-facing examples.
Audience and scenarios are well-covered across domains and tools with detailed docs. Capability boundaries are defined via skill categories and descriptions, but limitations per skill are not explicit. Trigger precision is supported by SKILL.md and commands, but trigger conditions are not specified. Environment fit is strong with multi-tool support and conversion scripts.
Information architecture is clear with organized directories and README/CHANGELOG. Install notes are detailed with multiple methods. Naming stability is supported by semantic versioning, but naming conventions are not explicit. Examples and FAQ are abundant. Known limitations are not listed. License is MIT. Versioning and changelog exist. Maintenance responsibility is indicated by CONTRIBUTING.md and CI, but maintainers are not clearly identified.
Output usability is plausible with structured instructions and scripts, but not verified. Marginal value is high given the breadth of skills. Cost-benefit is claimed (zero dependencies) but not validated.
Claim traceability is partial; third-party sources are mentioned but not linked. Cross-source corroboration is absent. Fact-inference separation is not explicitly marked, but README distinguishes facts and inferences to some extent.
- The repository is large, but static review cannot verify the actual functionality and quality of all skills.
- Inconsistency in skill counts (362 vs 345) in README may affect user expectations.
- No explicit least-privilege or user confirmation mechanisms; users must review skill behavior themselves.
- Dependency security audit only covers requirements files, not all scripts.
- Known limitations are not listed, which may lead to misunderstandings about skill boundaries.
What does this agent do, and when should you use it?
This is a large open-source library of 362 skills, 102 agents, 116 commands, and 644 Python CLI tools (all stdlib-only) for 13 AI coding tools, including Claude Code, OpenAI Codex, Gemini CLI, Cursor, and others. Skills are structured as folders with a SKILL.md, optional scripts/, references/, and assets/. The library covers diverse domains: engineering, DevOps, marketing (including AEO for answer-engine optimization), security, compliance, C-level advisory, productivity, academic research, and research operations. It installs as a Claude Code plugin marketplace, and can be converted to other coding agents via scripts. It includes 7 pre-configured personas (e.g., Startup CTO, Growth Marketer, Solo Founder) and an orchestration protocol for multi-agent workflows.
The repository provides installable skill bundles, each a folder with SKILL.md (instructions), scripts/ (Python tools), references/ (templates and domain knowledge). In Claude Code, users add the marketplace with /plugin marketplace add alirezarezvani/claude-skills, then install by domain with /plugin install <bundle>@claude-code-skills. For other tools, it runs conversion scripts (e.g., convert.sh --tool all) to generate native formats (e.g., .mdc files for Cursor), then installs via install.sh. Specific operations include: python3 finance/saas-metrics-coach/scripts/metrics_calculator.py --mrr 80000 --customers 200 --churned 3 --json computes SaaS metrics; brand_voice_analyzer.py analyzes brand voice; tech_debt_analyzer.py scores tech debt in a codebase; rice_prioritizer.py prioritizes features by RICE; skill_security_auditor.py scans skills for malicious code. It includes a PreToolUse hook for security guidance. For Hermes Agent and Mistral Vibe, it ships pre-generated skill trees with sync scripts (sync-hermes-skills.py, vibe-install.sh).
- Software engineering teams wanting to add domain-specific skills like architecture review, security auditing, or CI/CD generation to Claude Code.
- Developers using OpenAI Codex or Cursor who want the same skill set in native formats (e.g., .mdc rules).
- Marketing teams looking for SEO skills including AEO (E-E-A-T audits for LLM citations) and local SEO management.
- Solo founders needing personas like startup-cto and productivity skills like weekly-review and deep-work to streamline their work.
- Regulated industries (e.g., medical devices, finance) requiring compliance skills (MDR, ISO 27001, SOC 2) to review documentation and processes.
What are this agent's strengths and limitations?
- Unmatched skill coverage for multiple tools (362 skills across 18 domains), reducing the need for multiple libraries.
- All 644 Python tools are stdlib-only with zero pip dependencies, simplifying deployment and auditing.
- Includes 7 pre-configured personas and an orchestration protocol for advanced multi-agent workflows.
- Dedicated security auditor (skill-security-auditor) scans skills for malicious code, providing safety before installation.
- The library is massive, which can be overwhelming for users who only need a few skills.
- Conversion scripts (convert.sh) depend on Python and shell environments, which may not be native to all IDEs.
- For tools that don't support SKILL.md standard, compatibility requires conversion and installation steps, which may not be perfect.
- Some skills like C-level advisory are domain-specific and may require specific context, not applicable to general coding tasks.
How do you install or deploy this agent?
For Claude Code:
/plugin marketplace add alirezarezvani/claude-skills
/plugin install engineering-skills@claude-code-skillsFor other tools (e.g., Cursor or Aider):
git clone https://github.com/alirezarezvani/claude-skills.git
cd claude-skills
./scripts/convert.sh --tool all
./scripts/install.sh --tool cursor --target .For OpenAI Codex: npx agent-skills-cli add alirezarezvani/claude-skills --agent codex. Manual install: copy any skill folder to ~/.claude/skills/ or ~/.codex/skills/.
How do you use this agent?
Once installed, invoke skills by name in your supported coding agent. For example, in Claude Code, prompt "Using the senior-architect skill, review our microservices architecture and identify the top 3 scalability risks." Or, in Gemini CLI, use activate_skill(name="senior-architect"). For Python tools, run scripts directly, e.g., python3 product-team/product-manager-toolkit/scripts/rice_prioritizer.py features.csv. For security auditing, run python3 engineering/skill-security-auditor/scripts/skill_security_auditor.py /path/to/skill/.
FAQ
Is there a cost to use this library?
Can these skills be used with any programming language?
Are there conflicts when installing multiple skill bundles?
How do I convert skills for tools like Cursor or Aider?
./scripts/convert.sh --tool all to generate formats for all tools (e.g., .mdc files, CONVENTIONS.md), then install with ./scripts/install.sh --tool <name> --target <path>.