[FEATURE] Skill Discovery & Duplicate Prevention

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

Problem Statement: Skill Discovery & Duplicate Prevention

Current State

When users create custom skills in Claude Code, there is no automated mechanism to discover existing skills in public marketplaces, repositories, or the Claude ecosystem. This leads to:

1. Duplicate effort - Users unknowingly recreate skills that already exist
2. Fragmentation - Multiple implementations of similar functionality across projects
3. Missed opportunities - Existing, battle-tested skills remain undiscovered
4. Poor discoverability - No central way to search for skills before building

User Pain Points

Problem Scenario

User: "Claude, create a skill to manage Jira tickets"
Claude: *Creates new skill*

Reality: A well-maintained jira-workflow skill already exists in the marketplace
Result: Wasted time, duplicate maintenance, missed features

Current Manual Process

1. User asks Claude to create a skill
2. User must manually search:

• GitHub repositories
• Claude marketplace (if exists)
• Community forums
• .claude/ config files

1. User must evaluate similarity and quality
2. User decides to use existing vs. create new
3. This entire process is manual and error-prone

Benefits

For Users

• ✅ Save time - Don't rebuild what exists
• ✅ Better quality - Use battle-tested, maintained skills
• ✅ Discover features - Learn what's possible through existing skills
• ✅ Community alignment - Use standardized approaches

For Ecosystem

• ✅ Reduce fragmentation - Consolidate around quality implementations
• ✅ Encourage contribution - Users improve existing skills vs. creating duplicates
• ✅ Better metrics - Track which skills are actually useful
• ✅ Quality signal - Popular skills rise to the top

For Claude Code Team

• ✅ Improve UX - Proactive suggestions reduce friction
• ✅ Ecosystem growth - Promotes skill marketplace adoption
• ✅ Reduce support - Users find existing solutions faster
• ✅ Data insights - Understand what users need

Technical Considerations

Required Components

1. Skill Registry API/MCP Server

• Indexes public skill repositories (GitHub, GitLab, etc.)
• Provides search/similarity matching endpoints
• Returns metadata: stars, last update, description, usage stats

1. Similarity Matching Algorithm

• Semantic search on skill descriptions/documentation
• Intent matching based on user's natural language request
• Category/tag-based filtering

1. Integration Points

• Hook into skill creation flow (/create-skill or Skill tool)
• Background indexing of known skill sources
• Cache/CDN for fast lookups

1. Skill Metadata Standard

• Description, category, tags
• Compatibility/dependencies
• Maintenance status, popularity metrics
• Installation instructions

Data Sources to Index

• GitHub topics: claude-skill, claude-code-skill, mcp-skill
• Official Claude marketplace (if exists)
• Community-maintained registries
• User-contributed skill directories
• Private enterprise registries (opt-in)

Privacy/Security Considerations

• Users can disable marketplace querying (offline mode)
• Private/enterprise registries require authentication
• No telemetry on user's skill creation without consent
• Sandboxed skill execution (already exists in Claude Code)

Success Metrics

• Skill Reuse Rate: % of times users install existing skill vs. create new
• Time to Skill: Average time from request to working skill (should decrease)
• Duplicate Reduction: Fewer functionally-identical skills created
• Marketplace Engagement: Increased skill marketplace usage/contributions
• User Satisfaction: Measured via surveys on skill discovery experience

Open Questions

1. Marketplace Ownership: Who maintains the central skill registry? Anthropic? Community? Both?
2. Quality Control: How do we prevent low-quality/malicious skills from ranking high?
3. Versioning: How do we handle skill versioning and breaking changes?
4. Private Skills: Should enterprises be able to host private skill registries?
5. Monetization: Should skill authors be able to monetize popular skills?

Alternatives Considered

Manual Registry File

Approach: Users maintain a local skills-registry.json with known skills
Pros: Simple, no external dependency
Cons: Manual maintenance, no community benefit, stale data

Git Submodules for Skills

Approach: Skills live as git submodules in project
Pros: Version control, standard Git workflow
Cons: Doesn't solve discovery problem, complex for users

LLM-Based Search Only

Approach: Claude searches GitHub/web in real-time for similar skills
Pros: No new infrastructure
Cons: Slow, unreliable, no quality signals, rate limits

Recommendation

Implement a phased approach:

Phase 1: Basic Discovery (MVP)

• MCP server that indexes GitHub topics
• Simple keyword matching on skill descriptions
• Show top 3 matches before skill creation
• Manual installation link

Phase 2: Smart Recommendations

• Semantic similarity matching (embeddings)
• Popularity/quality signals (stars, maintenance, usage)
• One-click installation
• Update notifications

Phase 3: Ecosystem Platform

• Official skill marketplace/registry
• Community ratings/reviews
• Skill analytics dashboard
• Private enterprise registries

Related Work

• NPM/PyPI - Package discovery before installation
• VS Code Extensions - Marketplace with search, ratings, recommendations
• Homebrew - brew search before creating formula
• Docker Hub - Image search before building custom

---

Submitted by: Engineering Team
Date: 2026-03-11
Priority: Medium-High (impacts developer experience & ecosystem growth)
Estimated Effort: Medium (requires new infrastructure but clear scope)

Proposed Solution

Proposed Solution

Automatic Skill Discovery

Before creating a new skill, Claude Code should:

1. Query skill registries/marketplaces for similar functionality
2. Show recommendations with similarity scores
3. Offer installation of existing skills vs. creating new ones
4. Learn from community - surface popular/well-maintained options

Example Flow

User: "Create a skill to deploy to AWS using CDK"

Claude: "I found 3 existing skills that might match:
  1. aws-cdk-deploy (★ 245, maintained, 2 weeks ago)
     - Handles CDK deployment with stack management
     - Similarity: 92%

  2. cdk-helper (★ 89, active, 1 month ago)
     - Basic CDK deployment automation
     - Similarity: 78%

Would you like to:
  [Install] one of these skills
  [Review] their documentation
  [Create new] skill anyway
"

Alternative Solutions

_No response_

Priority

Medium - Would be very helpful

Feature Category

CLI commands and flags

Use Case Example

Use Cases

1. Pre-Creation Discovery

Scenario: User asks Claude to create a skill for common task
Desired: Claude searches registries, shows existing options before creating new

2. Similarity Detection

Scenario: User starts creating skill, Claude detects overlap with existing skill
Desired: Mid-creation, Claude suggests: "This looks similar to X, would you like to use that instead?"

3. Skill Health Dashboard

Scenario: User has installed skills but doesn't know if better alternatives exist
Desired: Claude proactively suggests: "Skill X has a more maintained alternative Y"

4. Version Management

Scenario: Installed skill has updates or deprecated in favor of newer version
Desired: Claude notifies: "Update available for skill X" or "Skill X is deprecated, migrate to Y"

Additional Context

_No response_