AgentSkillsCN

analyze-agent-overlap

分析现有代理、技能、提示与指令,识别其中的重叠、冗余与冲突。可与 GitHub Copilot、Claude Code、Codex、OpenCode 等提供商协同合作。在创建新的自定义文件之前,可优先选用此技能,以避免重复;在整合代理时,亦可优先选用此技能;或在排查冲突行为时,可优先选用此技能。

SKILL.md
--- frontmatter
name: analyze-agent-overlap
description: Analyzes existing agents, skills, prompts, and instructions to identify overlaps, redundancies, and conflicts. Works with GitHub Copilot, Claude Code, Codex, OpenCode, and other providers. Use before creating new customization files to avoid duplication, when consolidating agents, or when troubleshooting conflicting behaviors.

Analyze Agent Overlap

Detects redundancy, overlap, and potential conflicts between AI coding assistant customization files.

Provider Folder Reference

This skill works across multiple AI coding assistant providers:

ProviderBase Folder
GitHub Copilot.github/
Claude Code.claude/
Codex.codex/
OpenCode.config/opencode/

Throughout this document, <provider>/ represents your chosen provider's base folder.

When to Use

  • Before creating a new agent, skill, prompt, or instruction
  • When you suspect two agents are doing similar things
  • To audit and consolidate your customization files
  • When agent behaviors seem to conflict

Analysis Process

Step 1: Inventory Existing Items

Scan these locations (replace <provider>/ with actual folder):

  • <provider>/agents/*.md - All agent definitions (including .subagent.agent.md)
  • <provider>/skills/*/SKILL.md - All skill definitions
  • <provider>/prompts/*.prompt.md - All prompt templates
  • <provider>/instructions/*.instructions.md - All instruction files

For each item, extract:

  • Name: The identifier
  • Purpose: What problem it solves (from description)
  • Domain: What areas/topics it covers
  • Triggers: Keywords or scenarios that activate it
  • User-Invokable: Whether it's a user-facing agent or sub-agent

Step 2: Compare Against Proposed Item

When analyzing a proposed new item, compare:

Direct Overlap Indicators:

  • Same or very similar name
  • Same primary purpose statement
  • Identical target domain
  • Overlapping trigger keywords (>50% match)

Partial Overlap Indicators:

  • Related but distinct purposes
  • Some shared expertise areas
  • Similar but different trigger scenarios
  • Complementary functionality

No Overlap Indicators:

  • Different domains entirely
  • Non-overlapping use cases
  • Distinct trigger keywords

Step 3: Detect Conflicts

Look for these conflict types:

Behavioral Conflicts:

  • Two agents giving contradictory guidance for same scenario
  • Instructions that override each other for same file patterns
  • Skills that produce incompatible outputs

Scope Conflicts:

  • Multiple agents claiming the same use cases
  • Overlapping applyTo patterns in instructions
  • Ambiguous routing between similar agents

Naming Conflicts:

  • Names too similar causing confusion
  • Same name in different contexts

Overlap Severity Levels

🔴 Critical (Do Not Proceed)

  • Exact duplicate of existing item
  • Direct contradiction with existing guidance
  • Name collision
  • 80% purpose overlap

🟡 Warning (Needs Discussion)

  • Significant overlap (50-80% shared purpose)
  • Potential user confusion about which to use
  • Overlapping triggers with different behaviors
  • Partial scope conflict

🟢 Low Risk (Proceed with Awareness)

  • Minor overlap (<50% shared concerns)
  • Complementary purposes
  • Clear differentiation possible
  • Different trigger contexts

Resolution Strategies

When overlap is detected, consider:

Merge

Combine into single, more comprehensive item.

  • Best when: Items serve nearly identical purpose
  • Action: Create unified item, deprecate duplicates

Extend

Add new functionality to existing item.

  • Best when: New need is subset of existing item's scope
  • Action: Modify existing item, don't create new

Differentiate

Clarify boundaries between items.

  • Best when: Items serve related but distinct needs
  • Action: Update descriptions to make distinctions clear

Reference

Have one item delegate to another.

  • Best when: Items have hierarchical relationship
  • Action: Add handoff or reference in description

Supersede

Replace older item with improved version.

  • Best when: New item is strictly better
  • Action: Create new, mark old as deprecated

Output Format

markdown
## Overlap Analysis: [Proposed Item Name]

### Summary
**Proposed Type:** [Agent|Skill|Prompt|Instruction]
**Proposed Purpose:** [Brief description]
**Overlap Level:** None | Low | Medium | High | Critical
**Recommendation:** Proceed | Modify | Merge | Reconsider

### Comparison Matrix

| Existing Item | Type | Overlap | Shared Concerns |
|---------------|------|---------|-----------------|
| [name] | [type] | [level] | [what overlaps] |

### Detailed Findings

#### High/Critical Overlap Items
[For each significant overlap:]

**[Existing Item Name]**
- Type: [type]
- Purpose: [their purpose]
- Overlap Areas: [specific shared concerns]
- Key Distinction: [how proposed differs]
- Resolution: [recommended action]

#### Potential Conflicts
[List any behavioral or scope conflicts]

#### Complementary Items
[Items that could work well alongside proposed]

### Recommendations

1. [Primary recommendation with rationale]
2. [Secondary options if applicable]

### Questions to Resolve
- [Clarifying questions that would help decision]

Example Analysis

Proposed: database-helper agent for SQL query assistance

Findings:

  • dx12-terrain-engine-dev - No overlap (different domain)
  • agent-builder - No overlap (different domain)

Result: ✅ Proceed - no conflicts detected


Proposed: code-reviewer agent for code review

Findings:

  • Existing dx12-terrain-engine-dev mentions code quality
  • Partial overlap in "review code" scenarios

Result: ⚠️ Warning - clarify scope boundaries

  • Recommendation: code-reviewer for general review, dx12-terrain-engine-dev for DX12-specific review only