Skip to content
Back to skills

Skill Combination Techniques

ASecurity

Stack and chain multiple skills for complex tasks.

  • 2 stars
  • 0 votes
  • 0 copies
  • 4 views
  • Added September 10, 2026
devopssqldockerkubernetestestingcode-reviewgitapibackend

Works with

  • api

Security analysis

A100/100

Scanned September 10, 2026

npx -y skills add LoopyLuci/Skills --skill skill-combination-techniques --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Skill Combination Techniques?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Skill Combination Techniques
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/loopyluci-skill-combination-techniques/badge)](https://www.skillsdirectory.com/skills/loopyluci-skill-combination-techniques)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: skill-combination-techniques
description: Stack and chain multiple skills for complex tasks.
---

# Skill Combination Techniques

**Trigger**: Use when a task requires expertise from multiple domains and needs multiple skills working together.

## When to Combine Skills

```markdown
COMBINE skills when:
- Task spans multiple domains (e.g., deploy + monitor + alert)
- Task has multiple phases (build → test → deploy)
- The skill references "Connected Skills" in its body

DON'T COMBINE when:
- One skill fully covers the task
- Skills overlap significantly (redundant)
- Combining would exceed context budget
```

## Combination Patterns

### 1. Sequential Chain (most common)
```
Phase 1: /github-actions-workflows    → Create CI pipeline
Phase 2: /dockerfile-optimization     → Optimize Docker build
Phase 3: /kubernetes-deployment       → Deploy to k8s
```
Each phase completes before the next loads. Spread across turns.

### 2. Parallel Stack (loaded together)
```
/docker-compose-patterns /github-actions-workflows /secrets-management
```
All needed simultaneously. Max 5 per stack.

### 3. Nested (skill within a skill)
```
Primary: /webapp-penetration-testing
  ├── Sub: /sql-injection-exploitation
  └── Sub: /api-testing-contracts
```
Load primary first. Only load sub-skills when you reach that section.

### 4. Fallback Chain
```
Primary:  /docker-compose-migration    → Latest approach
Fallback: /docker-compose-patterns    → Alternative if primary fails
```

## The Stack Command

```markdown
# Up to 5 leading /skill tokens, rest is instruction
/github-pr-workflow /test-driven-development fix issue #123

# Bundles are better for permanent combos
hermes bundles create backend-dev \
  --skill github-code-review \
  --skill test-driven-development
```

## Skill Handoff Protocol

When transitioning between skills in a chain:

1. **ANNOUNCE**: "Now using <next-skill> for <next-phase>"
2. **SUMMARIZE** what previous skill accomplished
3. **LOAD** next skill only when needed
4. **PASS context** between skills

```
Example:
"Using github-actions-workflows, I've set up the CI pipeline.
Now loading dockerfile-optimization for the Docker build step..."
```

## Connected Skills

When a skill lists "Connected Skills", load them only when you reach that subsection:

```
## Connected Skills
`dockerfile-optimization`, `helm-chart-development`, `prometheus-metrics`

Strategy:
1. Load kubernetes-deployment first
2. At "Build container" → load dockerfile-optimization
3. At "Deploy with Helm" → load helm-chart-development
4. At "Set up monitoring" → load prometheus-metrics
```

## Pitfalls
- **Premature loading**: Loading all phase skills upfront wastes context
- **Handoff failure**: Starting new skill without communicating what old skill did
- **Over-stacking**: >5 skills in one message — Hermes caps at 5
- **Bundle vs stack**: Same combo 3+ times → create a bundle instead

## Verification
```markdown
After multi-skill task:
- Did each skill add unique value?
- Was there overlap?
- Would a bundle be better next time?
```

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.