Skills

Skills teach the AI agent specific workflows and domain knowledge via SKILL.md files. Each skill is a structured markdown file that gets injected into the system prompt when enabled.

How Skills Work

When enabled, a skill's content is injected into the system prompt, giving the agent specialized knowledge for tasks like SwiftUI development, code review, or Kubernetes deployment. Skills include a persona, core expertise, patterns, best practices, anti-patterns, and verification criteria.

Skill Locations

60+ Bundled Skills

Swift/Apple

• Swift & iOS Development — SwiftUI, Apple frameworks, Xcode, HIG
• Async/Await Conversion — Combine and callback migration to modern Swift concurrency
• SwiftUI Migration — UIKit-to-SwiftUI migration patterns
• SwiftData Migration — Core Data to SwiftData migration
• Core ML Conversion — PyTorch/TensorFlow to Core ML format
• MLX Training — Fine-tuning models on Apple Silicon
• App Store Prep — Review guidelines, metadata, privacy manifests
• Privacy Manifest Audit — PrivacyInfo.xcprivacy compliance

Web & Frontend

• React & Next.js — Server Components, App Router, TypeScript
• Full-Stack Development — End-to-end web application architecture
• Flutter & Dart — Cross-platform mobile and desktop development

AI & Agents

• AI Agent Design — Autonomous agents with tool use, memory, and planning loops
• MCP Server Building — Model Context Protocol servers for AI integration
• RAG Pipeline Design — Retrieval-augmented generation with embeddings and vector search
• Fine-Tuning & Eval — LLM fine-tuning, dataset curation, and evaluation
• LLM Observability — Cost tracking, latency analysis, and quality monitoring
• Prompt Engineering — System prompts, few-shot patterns, tool descriptions

Security & Offensive

• Security Audit — OWASP coverage, vulnerability scanning, remediation
• Penetration Testing — Structured pentesting with proof-of-concept exploitation
• Exploit Analysis — CVE assessment, attack chains, mitigation strategies
• Network Forensics — Packet analysis, incident investigation, digital forensics
• Reverse Engineering — Binary analysis, protocol deconstruction, API exploration

Infrastructure & DevOps

• DevOps — CI/CD pipelines, Docker, deployment infrastructure
• Kubernetes — Deployments, services, operational patterns
• Terraform IaC — Infrastructure as code, modules, state management
• AWS Serverless — Lambda, API Gateway, DynamoDB, CDK
• Docker & Container Deployment — Containerization and orchestration
• CI/CD Pipeline Design — GitHub Actions, matrix builds, artifact management
• Edge Computing — CDN workers, edge functions, distributed caching
• Observability — Structured logging, distributed tracing, metrics, alerting
• Incident Response — Triage, communication, resolution, post-mortems
• Platform Engineering — Internal developer platforms, golden paths, DevEx
• Cloud Cost Optimization — Right-sizing, reserved capacity, FinOps

Architecture & Design

• System Design — Distributed systems, scalability, reliability patterns
• API Design — REST/GraphQL conventions, versioning, error handling
• Database Design — Schema modeling, migrations, query optimization
• GraphQL Design — Schema design, resolvers, subscriptions, security
• Monorepo Management — Multi-package repos, build systems, versioning

Development Practices

• Code Review — Systematic code review methodology
• PR Code Review — Pull request review with structured feedback
• Testing — Unit, integration, and E2E test design
• Test Generation — Auto-generating test suites from code
• Debugging — Systematic root cause analysis
• Refactoring — Safe code transformation patterns
• Performance — Profiling, benchmarking, optimization
• Documentation — Technical docs, API references, guides
• Technical Writing — RFCs, ADRs, developer guides
• Code Migration — Framework and language migration strategies

Business & Strategy

• Competitive Analysis — Market positioning and differentiation
• Competitive Intelligence — Systematic competitor research and monitoring
• Product Strategy — Vision, roadmaps, and prioritization
• Pricing & Monetization — Pricing models, tiers, monetization strategy
• Growth Analytics — Funnels, experimentation, retention analysis
• Pitch Deck Review — Fundraising deck evaluation
• Technical Due Diligence — Codebase and team assessment for M&A

General

• Research — Systematic information gathering and synthesis
• Writing — Clear, effective technical and business writing
• Planning — Project scoping, breakdown, and risk assessment
• Specification — Requirements gathering and spec writing
• Argumentation — Structured reasoning and argument construction
• Rapid Prototyping — MVP scaffolding and validation
• Regex Expert — Pattern design across engines
• Data Science — ML pipelines, statistics, visualization

Languages & Frameworks

• Rust Systems — Ownership, concurrency, performance
• Unity Game Dev — C# scripting, engine best practices

Meta-Skills (Combos)

• Architect Mode — System design + API design + database + security
• Ship It Mode — Rapid prototyping + full-stack + testing + DevOps
• War Room Mode — Incident response + debugging + performance + observability
• Deep Dive Mode — Research + documentation + code review + testing
• Teacher Mode — Research + writing + documentation + examples
• Red Team Mode — Security audit + pentesting + exploits + reverse engineering

SKILL.md Format

Every skill follows the elite template:

---
name: Skill Name
description: One-line description of what this skill enables.
tags: [tag1, tag2, tag3]
---
 
You are an expert [role] who [specializes in what].
 
## Core Expertise
- Bullet list of domain knowledge areas
 
## Patterns & Workflow
1. **Step** — Description of each step in the workflow
 
## Best Practices
- Actionable guidelines for high-quality work
 
## Anti-Patterns
- Common mistakes to avoid with explanation of why
 
## Verification
- Checkable criteria to confirm work quality
 
## Examples
- **Scenario**: Concrete example showing the skill in action

ClawHub

ClawHub is a shared skill registry between G-Rump and OpenClaw. Skills installed via ClawHub are stored in ~/.grump/skills/ and accessible from both systems.

Browse and manage hub skills in Settings → Workspace → Skills → ClawHub.

Managing Skills

• Settings → Workspace → Skills — Toggle skills on/off, add/edit/remove
• Skills can be enabled/disabled per-conversation via workflow presets
• The agent sees only enabled skills in its system prompt
• Meta-skills (combos) activate multiple skills simultaneously for complex workflows