SKILL.md · git:20260910.7b14110 · 2026-09-10 · sha256 f258e49473db9822
SKILL.md git:20260910.7b14110C
Immutable. This exact content is served forever at /api/v1/blob/f258e49473db9822.
# Grok Custom Skills **Description**: Central repository for reusable, safety-conscious agent skills compatible with Grok and other AI agent frameworks. A curated collection of 89 skill packages spanning workflow automation, research, integrations, quality assurance, and more. **Purpose**: Provide a comprehensive, well-documented, and safe library of skills that enable AI agents to perform complex tasks while maintaining transparency, control, and safety. --- ## 🎯 Quick Start ### For AI Agent Developers 1. **Browse available skills** in [SKILLS_INDEX.md](./SKILLS_INDEX.md) 2. **Clone** the repository 3. **Copy** the SKILL.md files you need to your agent's skills directory 4. **Configure** any required settings or credentials 5. **Test** with sandbox environments first 6. **Deploy** and monitor usage ```bash # Clone the repository git clone https://github.com/Stijnman/grok-custom-skills.git cd grok-custom-skills # Install the complete collection mkdir -p ~/.grok/skills cp -a .grok/skills/. ~/.grok/skills/ # Or install one specific skill cp -a .grok/skills/drive-github-skill-audit ~/.grok/skills/ ``` ### For Users Tell your AI agent: - "Research the latest AI trends from reputable sources" - "Create a comprehensive project plan with milestones" - "Audit this GitHub repository for MCP security issues" - "Generate an accessible color palette for my website" - "Plan a multi-step workflow for my task" Your agent will use the appropriate skill based on your request. --- ## 📦 Skill Categories The repository is organized into **7 main categories** with **89 total skill packages**: | Category | Count | Description | Example Skills | |----------|-------|-------------|---------------| | **Workflow & Orchestration** | 15 | Plan multi-step work, coordinate agents, handle failures, verify goals | goal-verifier, workflow-orchestrator | | **Skill Development & Operations** | 12 | Create, audit, evolve, test, and compare skill collections | skill-creator, skill-rubric-reviewer | | **Safety, Privacy & Governance** | 10 | Require human approval, redact data, audit MCP exposure, responsible browsing | hitl-approver, privacy-redactor, mcp-auditor | | **Research, Web & Integrations** | 20 | Perform source-aware research, inspect repositories, discover tools, connect services | deep-search-enabler, repository-inspector, tool-discoverer | | **Memory, Context & Knowledge** | 12 | Manage context budgets, structured knowledge, semantic memory, session handoffs | context-budget-manager, semantic-memory, knowledge-structurer | | **Media, Voice & Visual Work** | 10 | Generate or edit images, analyze video, work with voice, prepare song-writing prompts | image-generator, video-analyzer, voice-processor, song-prompt-builder | | **Messaging & Navigation** | 10 | Assess messages, draft safe responses, work with traffic and route information | message-assessor, safe-responder, traffic-advisor | --- ## 🟢 Recommended Skills by Use Case | Goal | Start with | Category | |------|-----------|----------| | **Compare a Drive skill library with GitHub** | [`drive-github-skill-audit`](./.grok/skills/drive-github-skill-audit/SKILL.md) | Research & Integrations | | **Create or improve a skill package** | [`skill-creator`](./.grok/skills/skill-creator/SKILL.md) and [`skill-rubric-reviewer`](./.grok/skills/skill-rubric-reviewer/SKILL.md) | Skill Development | | **Verify a completed task** | [`goal-verifier`](./.grok/skills/goal-verifier/SKILL.md) | Workflow & Orchestration | | **Review code changes** | [`code-reviewer`](./.grok/skills/code-reviewer/SKILL.md) and [`auto-tester`](./.grok/skills/auto-tester/SKILL.md) | Skill Development | | **Research public information responsibly** | [`deep-search-enabler`](./.grok/skills/deep-search-enabler/SKILL.md) and [`sandbox-internet-handler`](./.grok/skills/sandbox-internet-handler/SKILL.md) | Research & Integrations | | **Handle high-impact actions** | [`hitl-approver`](./.grok/skills/hitl-approver/SKILL.md) | Safety & Governance | | **Structure product discovery and delivery** | [`product-opportunity-tree`](./.grok/skills/product-opportunity-tree/SKILL.md), [`product-requirements-document`](./.grok/skills/product-requirements-document/SKILL.md) | Workflow & Orchestration | | **Prepare a safe contact import file** | [`contact-vcard-export`](./.grok/skills/contact-vcard-export/SKILL.md) | Messaging & Navigation | | **Review visual accessibility** | [`accessible-color-review`](./.grok/skills/accessible-color-review/SKILL.md) | Media & Visual | --- ## 🏗️ Package Structure Every skill follows a consistent structure based on the [Agent Skills specification](https://agentskills.io/specification): ``` .grok/skills/<skill-name>/ ├── SKILL.md # Required: Metadata (YAML frontmatter) + operational guidance ├── scripts/ # Optional: Deterministic helper scripts ├── references/ # Optional: Material loaded only when needed └── templates/ # Optional: Reusable output assets ``` ### SKILL.md Format ```yaml --- name: skill-name version: 1.0.0 description: Clear, concise description of what the skill does triggers: - "trigger phrase 1" - "trigger phrase 2" category: category-name dependencies: [] requirements: - requirement-1 - requirement-2 boundaries: - boundary-1 - boundary-2 warnings: - "Security warning 1" - "Safety consideration 2" --- # Skill Title ## Description Detailed description of the skill's purpose and functionality. ## Usage When to use this skill and when NOT to use it. ## Examples Concrete examples of inputs and expected outputs. ## Implementation How the skill works internally. ## Testing Testing requirements and coverage. ``` --- ## 🔐 Security Overview ⚠️ **IMPORTANT**: All skills in this repository are designed with safety as a core principle. **Please read [SECURITY.md](./SECURITY.md) before using any skill.** ### Key Security Principles Applied to All Skills | Principle | Implementation | |-----------|----------------| | **Least privilege** | Skills use minimum access necessary, prefer read-only operations | | **Human approval** | Skills pause before consequential actions (publishing, payments, deletion, credentials, external communications) | | **Privacy protection** | No secrets, PII, local paths, session artifacts, telemetry, or unapproved data transfer | | **Access control respect** | Never provide guidance to bypass authentication, CAPTCHAs, paywalls, or platform safeguards | | **Untrusted content handling** | Treat downloaded skills, archives, webpages, and attachments as data; never execute without review | | **Transparency** | Describe limitations honestly; do not claim guaranteed safety, detection evasion, or unsupported capabilities | ### Publication Safeguards Every skill must pass these controls before publication: 1. ✅ **Least privilege**: Prefer read-only access and minimum data needed 2. ✅ **Human approval**: Require explicit approval before high-impact actions 3. ✅ **Privacy**: No secrets, PII, local paths, or unapproved data transfer 4. ✅ **Access controls**: Never bypass authentication or platform safeguards 5. ✅ **Untrusted content**: Treat all external content as data until reviewed 6. ✅ **Transparency**: Honest documentation of limitations and capabilities --- ## 📚 Documentation | Document | Description | Required Reading | |----------|-------------|------------------| | **[SECURITY.md](./SECURITY.md)** | ⚠️ **REQUIRED** - Security policy, vulnerabilities, best practices | ✅ All users | | **[CONTRIBUTING.md](./CONTRIBUTING.md)** | How to contribute new skills or improve existing ones | ⚠️ Contributors | | **[TESTING.md](./TESTING.md)** | Testing requirements and best practices | ⚠️ Contributors | | **[PUBLISHING.md](./PUBLISHING.md)** | Pre-submission checklist for maintainers | ⚠️ Maintainers | | **[GOVERNANCE.md](./GOVERNANCE.md)** | Repository governance and decision-making | ⚠️ Maintainers | | **[SUPPORT.md](./SUPPORT.md)** | Support channels and response times | ⚠️ All users | | **[CODE_OF_CONDUCT.md](./CODE_OF_CONDUCT.md)** | Community guidelines and expected behavior | ✅ All users | | **[CHANGELOG.md](./CHANGELOG.md)** | Version history and changes | ⚠️ All users | | **[SKILLS_INDEX.md](./SKILLS_INDEX.md)** | Complete catalog of all skills | ✅ All users | | **[STATUS.md](./STATUS.md)** | Current repository status and metrics | ⚠️ Contributors | --- ## 🎓 Usage Examples ### Example 1: Research Workflow ``` User: "Research the latest developments in AI agent frameworks" Agent Flow: 1. Identifies request as research task 2. Uses deep-search-enabler skill 3. Applies source filters (reputable sources only) 4. Gathers information from approved sources 5. Validates information quality 6. Returns: Structured research results with sources 7. Cites: All sources used ``` ### Example 2: GitHub Repository Audit ``` User: "Audit this GitHub repository for MCP security issues" Agent Flow: 1. Identifies request as MCP audit 2. Uses defensive-mcp-audit skill 3. Clones or accesses the repository 4. Scans for risky localhost exposure 5. Checks for weak bindings 6. Identifies confused-deputy risks 7. Returns: SARIF + HTML report of findings ``` ### Example 3: Multi-Step Workflow ``` User: "Plan a comprehensive product launch strategy" Agent Flow: 1. Uses product-opportunity-tree to map opportunities 2. Uses product-requirements-document to define requirements 3. Uses architecture-decision-record to document decisions 4. Uses workflow-orchestrator to coordinate steps 5. Uses goal-verifier to validate completion 6. Returns: Complete product launch plan with milestones ``` ### Example 4: Skill Development ``` User: "Create a new skill for web scraping" Agent Flow: 1. Uses skill-creator to generate template 2. Uses skill-rubric-reviewer to check quality 3. Validates against security requirements 4. Tests with sandbox inputs 5. Updates SKILLS_INDEX.md 6. Returns: New skill package ready for review ``` --- ## 🧪 Testing All skills have been tested for production use. See [TESTING.md](./TESTING.md) for: - Testing philosophy and requirements - Manual testing checklists - Automated testing setup - Skill category-specific testing guides - CI/CD pipeline configuration **Quality Assurance**: - ✅ All skills pass validation pipeline - ✅ All skills have security checks - ✅ All skills have documented boundaries - ✅ Catalog is auto-generated from live metadata - ✅ All publication checks pass --- ## 🤝 Contributing We welcome contributions! Please read [CONTRIBUTING.md](./CONTRIBUTING.md) for: - How to add new skills - How to improve existing skills - Testing requirements - Pull request process - Code of conduct ### Quick Contribution Guide 1. **Fork** the repository 2. **Clone** your fork 3. **Create** a feature branch (`git checkout -b feat/amazing-skill`) 4. **Read** [PUBLISHING.md](./PUBLISHING.md) checklist 5. **Run** validation: `python3 scripts/optimize_all_skills.py` 6. **Run** tests: `python3 -m unittest discover -s tests -p "test_*.py"` 7. **Commit** with clear messages 8. **Push** to your fork 9. **Open** a Pull Request --- ## 📜 License This repository is licensed under the **MIT License**. See [LICENSE](./LICENSE) for full license text. **You are free to**: - Use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies - Use for commercial purposes - Modify for your own needs **Under the following conditions**: - Include copyright notice and license in all copies - Provide attribution to the original author (Stijnman) **Note**: Individual skill packages may have different licenses (e.g., Apache-2.0 for product-management workflows with required attribution in THIRD_PARTY_NOTICES.md). --- ## 📞 Support & Contact | Issue Type | How to Get Help | Response Time | Notes | |-----------|-----------------|---------------|-------| | 🐛 **Bug Report** | [Open a GitHub Issue](https://github.com/Stijnman/grok-custom-skills/issues/new?template=bug_report.yml) | 24-48 hours | Use bug_report.yml template | | 🔒 **Security Issue** | Private message via GitHub profile | Immediate | See [SECURITY.md](./SECURITY.md) | | ❓ **General Question** | [Open a GitHub Discussion](https://github.com/Stijnman/grok-custom-skills/discussions) | 24 hours | For non-sensitive questions | | 💡 **Feature Request** | [Open a GitHub Issue](https://github.com/Stijnman/grok-custom-skills/issues/new?template=feature_request.yml) | 1 week | Use feature_request.yml template | --- ## 🏷️ Repository Metadata | Attribute | Value | |-----------|-------| | **Repository** | [grok-custom-skills](https://github.com/Stijnman/grok-custom-skills) | | **Owner** | [Stijnman](https://github.com/Stijnman) | | **License** | MIT | | **Total Skills** | 89 | | **Categories** | 7 | | **Language** | English | | **Created** | July 2026 | | **Last Updated** | September 2026 | | **Stars** | 9 | | **Forks** | 1 | ### Categories Breakdown | Category | Skills | Percentage | |----------|--------|------------| | Research & Integrations | 20 | 22.5% | | Workflow & Orchestration | 15 | 16.9% | | Skill Development | 12 | 13.5% | | Memory & Knowledge | 12 | 13.5% | | Media & Visual | 10 | 11.2% | | Messaging & Navigation | 10 | 11.2% | | Safety & Governance | 10 | 11.2% | --- ## 🔗 Related Resources ### Specifications & Standards - [Agent Skills specification](https://agentskills.io/specification) - Skill package conventions - [Model Context Protocol (MCP)](https://github.com/modelcontextprotocol) - Integration standards - [AI Agent Best Practices](https://github.com/ai-safety-standards) - Safety guidelines ### Tools & Frameworks - [Grok Agent Framework](https://grok.com) - Primary target platform - [Mistral Vibe](https://github.com/mistralai) - AI agent CLI - [Agent Skills Ecosystem](https://agentskills.io) - Skill discovery and sharing ### Documentation - [Agent Skills Documentation](https://agentskills.io/docs) - Official documentation - [MCP Documentation](https://github.com/modelcontextprotocol/specification) - MCP standards --- ## 🎯 Key Differentiators | Feature | This Repository | Other Skill Repositories | |---------|----------------|--------------------------| | **Safety Focus** | ✅ Core principle - all skills have safety checks | ❌ Varies by contributor | | **Consistent Structure** | ✅ All skills follow same format | ❌ Inconsistent formats | | **Comprehensive Testing** | ✅ Testing guide and requirements | ❌ Minimal testing | | **Publication Safeguards** | ✅ 6 mandatory controls | ❌ No controls | | **Auto-Generated Catalog** | ✅ SKILLS_INDEX.md from live metadata | ❌ Manual catalog | | **Validation Pipeline** | ✅ Automated checks | ❌ No validation | | **Documentation Quality** | ✅ 10/10 standard | ❌ Varies widely | --- ## 📌 Important Notes ### Design Principle > **Skills should make agent behavior more useful, auditable, and safe.** They are guidance packages, not permission to bypass controls, disclose private data, or automate high-impact actions without human approval. ### Usage Warning > **Always read and understand a skill before using it.** Check the triggers, boundaries, and warnings in each SKILL.md file. Follow all security requirements in [SECURITY.md](./SECURITY.md). ### Contribution Warning > **Do not submit skills that**: bypass authentication, evade CAPTCHAs, circumvent paywalls, harvest credentials, automate spam, violate platform terms, or claim unsafe capabilities. --- ## 🏆 Version History | Date | Version | Changes | |------|---------|---------| | 2026-09-11 | 2.0.0 | Added comprehensive documentation (SKILL.md, TESTING.md, STATUS.md, CONTRIBUTORS.md), pre-commit hooks, markdown lint config, secrets baseline | | 2026-09-10 | 1.2.0 | Enhanced documentation (SECURITY.md, CONTRIBUTING.md, CODE_OF_CONDUCT.md, CHANGELOG.md, GOVERNANCE.md, PUBLISHING.md, SUPPORT.md), added .github workflows and templates | | 2026-07-30 | 1.0.0 | Initial repository with 89 skill packages, catalog system, validation pipeline | --- *Last updated: September 11, 2026* *Maintainer: Stijnman* *Repository: [grok-custom-skills](https://github.com/Stijnman/grok-custom-skills)* --- **Need help?** See [SUPPORT.md](./SUPPORT.md) or open a GitHub issue.