Skip to content
Back to skills

Knowledge Discovery

ASecurity

Discover relevant skills and knowledge using tiered approach. SKIP for simple tasks, QUICK for single-skill, STANDARD for multi-skill, DEEP for full feature implementation. Auto-selects tier based on task complexity.

  • 666 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added February 9, 2026
toolsdockertestingdebugginggitapidatabasesecurity

Works with

  • api

Security analysis

A100/100

Scanned February 12, 2026

npx -y skills add majiayu000/claude-skill-registry --skill knowledge-discovery --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Knowledge Discovery?

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

Security grade badge for Knowledge Discovery
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/majiayu000-knowledge-discovery/badge)](https://www.skillsdirectory.com/skills/majiayu000-knowledge-discovery)

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: knowledge-discovery
description: "Discover relevant skills and knowledge using tiered approach. SKIP for simple tasks, QUICK for single-skill, STANDARD for multi-skill, DEEP for full feature implementation. Auto-selects tier based on task complexity."
layer: 4
tech_stack: [agnostic]
topics: [discovery, index, skills, knowledge, flows, context7, efficiency]
depends_on: []
complements: []
keywords: [discover, find, search, related, skills, knowledge, workflow, flow, tier, quick]
---

# Knowledge Discovery

Adaptive discovery skill with tiered efficiency for different task complexities.

## Discovery Tiers

| Tier | When to Use | Files Read | Time Cost |
|------|-------------|------------|-----------|
| **SKIP** | Simple/obvious tasks, single file edits | 0 | ~0s |
| **QUICK** | Single-skill tasks, known patterns | Quick ref only | ~2s |
| **STANDARD** | Multi-skill tasks, unfamiliar patterns | INDEX + GRAPH | ~10s |
| **DEEP** | Full feature implementation, CRUD flows | All 4 steps | ~15-20s |

## Tier Selection Guide

```
START: What's the task?
│
├─ "Fix typo" / "Add comment" / "Simple edit"
│   └─ SKIP (no discovery needed)
│
├─ "Write tests" / "Add validation" / "Single pattern"
│   └─ QUICK (use inline reference below)
│
├─ "Design API" / "Debug error" / "Multi-step task"
│   └─ STANDARD (read SKILL-INDEX + CONTEXT-GRAPH)
│
└─ "Implement CRUD" / "Add feature" / "Full workflow"
    └─ DEEP (full 4-step protocol)
```

## QUICK Tier: Inline Reference

**Use this for single-skill tasks without reading any files:**

| Task | Skill | No Lookup Needed |
|------|-------|------------------|
| Entity/DTO/AppService | `abp-framework-patterns` | ✓ |
| DbContext/Migration | `efcore-patterns` | ✓ |
| Input validation | `fluentvalidation-patterns` | ✓ |
| Permissions/Auth | `openiddict-authorization` | ✓ |
| Unit/Integration tests | `xunit-testing-patterns` | ✓ |
| E2E tests | `e2e-testing-patterns` | ✓ |
| Query optimization | `linq-optimization-patterns` | ✓ |
| API design | `api-design-principles` | ✓ |
| Technical design doc | `technical-design-patterns` | ✓ |
| Debug/errors | `debugging-patterns` | ✓ |
| Docker/.NET | `docker-dotnet-containerize` | ✓ |
| Git advanced | `git-advanced-workflows` | ✓ |
| Security audit | `security-patterns` | ✓ |

**Common Error → Skill:**

| Error | Skill |
|-------|-------|
| N+1 query | `linq-optimization-patterns` |
| Authorization failed | `openiddict-authorization` |
| Validation failed | `fluentvalidation-patterns` |
| DbUpdateException | `efcore-patterns` |
| Task was canceled | `dotnet-async-patterns` |

## STANDARD Tier Protocol

For multi-skill tasks, read these files:

```
1. DISCOVER  →  Check SKILL-INDEX.md for relevant skills
2. RELATE    →  Check CONTEXT-GRAPH.md for dependencies
```

## DEEP Tier Protocol

For full feature implementation:

```
1. DISCOVER  →  Check SKILL-INDEX.md for relevant skills
2. RELATE    →  Check CONTEXT-GRAPH.md for dependencies
3. REFERENCE →  Read /knowledge/ for shared patterns
4. FOLLOW    →  Use /flows/ for multi-step workflows
```

## Discovery by Task Type

### Creating a New Entity/Feature

1. **Check flow**: [flows/crud-implementation.md](../../flows/crud-implementation.md)
2. **Primary skills**: `abp-framework-patterns`, `efcore-patterns`, `fluentvalidation-patterns`
3. **Knowledge**:
   - [knowledge/entities/base-classes.md](../../knowledge/entities/base-classes.md)
   - [knowledge/examples/crud-entity.md](../../knowledge/examples/crud-entity.md)

### Adding Validation

1. **Primary skill**: `fluentvalidation-patterns`
2. **Knowledge**:
   - [knowledge/examples/validation-chain.md](../../knowledge/examples/validation-chain.md)
   - [knowledge/conventions/naming.md](../../knowledge/conventions/naming.md)

### Implementing Authorization

1. **Primary skill**: `openiddict-authorization`
2. **Related**: `security-patterns`, `abp-framework-patterns`
3. **Knowledge**: [knowledge/conventions/permissions.md](../../knowledge/conventions/permissions.md)

### Writing Tests

1. **Primary skill**: `xunit-testing-patterns`
2. **Knowledge**:
   - [knowledge/conventions/folder-structure.md](../../knowledge/conventions/folder-structure.md)
   - [knowledge/examples/crud-entity.md](../../knowledge/examples/crud-entity.md)

### Debugging an Error

1. **Check index**: Search error message in [SKILL-INDEX.md](../../SKILL-INDEX.md) (Error Message section)
2. **Primary skill**: `debugging-patterns`
3. **Related**: Check skill by keyword match

### Designing an API

1. **Primary skills**: `api-design-principles`, `technical-design-patterns`
2. **Related**: `domain-modeling`, `requirements-engineering`

### Optimizing Queries

1. **Primary skill**: `linq-optimization-patterns`
2. **Related**: `efcore-patterns`
3. **Check index**: Search "N+1" in [SKILL-INDEX.md](../../SKILL-INDEX.md)

## Skill Layer Loading

Load skills in dependency order:

```
Layer 1 (Foundations) → Load first
    csharp-advanced-patterns, dotnet-async-patterns, error-handling-patterns

Layer 2 (Framework) → Load second
    abp-framework-patterns, efcore-patterns, fluentvalidation-patterns

Layer 3 (Features) → Load third
    xunit-testing-patterns, security-patterns, distributed-events

Layer 4 (Workflows) → Load as needed
    feature-development-workflow
```

## Index Files Reference

| File | Purpose | Location |
|------|---------|----------|
| **SKILL-INDEX.md** | Find skills by task/keyword/error | [SKILL-INDEX.md](../../SKILL-INDEX.md) |
| **CONTEXT-GRAPH.md** | Skill dependencies, layers, relationships | [CONTEXT-GRAPH.md](../../CONTEXT-GRAPH.md) |
| **knowledge/INDEX.md** | Shared patterns and conventions | [knowledge/INDEX.md](../../knowledge/INDEX.md) |
| **flows/INDEX.md** | Multi-step workflows | [flows/INDEX.md](../../flows/INDEX.md) |

## Quick Lookup Commands

### By Task
```
"I need to create an entity" → abp-framework-patterns + efcore-patterns
"I need to validate input" → fluentvalidation-patterns
"I need to add permissions" → openiddict-authorization
"I need to write tests" → xunit-testing-patterns
"I need to optimize queries" → linq-optimization-patterns
```

### By Error
```
"N+1 query" → linq-optimization-patterns
"Authorization failed" → openiddict-authorization
"Validation failed" → fluentvalidation-patterns
"DbUpdateException" → efcore-patterns
```

### By Keyword
```
"Entity" → abp-framework-patterns
"DbContext" → efcore-patterns
"FluentValidation" → fluentvalidation-patterns
"[Authorize]" → openiddict-authorization
"xUnit" → xunit-testing-patterns
```

## Integration

This skill should be invoked:
- Automatically at the start of complex tasks
- When the user asks "what skills do I need for..."
- Before implementing multi-step workflows
- When encountering unfamiliar patterns or errors

## Related Skills

After discovery, typically load:
- `abp-framework-patterns` - Core ABP patterns
- `efcore-patterns` - Database layer
- `fluentvalidation-patterns` - Input validation
- `openiddict-authorization` - Permissions
- `xunit-testing-patterns` - Testing

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.