What do the similarity scores represent?

Scores range from 0 to 1; a score above 0.7 indicates a strong match, while scores below 0.5 suggest you should refine your query or use keyword search.

Can I use this on remote servers?

Yes, the skill is designed to work both locally and remotely, with specific support for accessing self-hosted services over Tailscale.

How does semantic search differ from standard grep?

While grep searches for literal text strings, semantic search understands the underlying concept and behavior of the code, finding relevant logic even if your search terms don't match the variable names.

Does this skill require external infrastructure?

Yes, it requires Qdrant and an embedding service running locally or via Docker to handle the vector indexing and GPU-accelerated search.

Semantic Code Explorer

Name: Semantic Code Explorer
Author: adilasif

byadilasif

0•

Developer Tools

Searches and explores codebases using GPU-accelerated semantic embeddings to find logic by intent rather than keywords.

Semantic Code Explorer transforms how developers navigate large or unfamiliar repositories by enabling search based on meaning and behavior instead of literal strings. By leveraging self-hosted vector search via Qdrant and Docker, this skill allows Claude to answer complex architectural questions like 'how is user state managed' or 'where is the retry logic handled' with high precision. It serves as a powerful alternative to traditional grep, providing relevancy scores and contextual snippets that help developers trace execution flows and understand codebase architecture faster.

Key Features

010 GitHub stars

02GPU-accelerated semantic search for rapid code indexing and retrieval

03Relevancy scoring to filter strong matches from noise

04Tailscale support for searching remote codebases securely

05Intent-based querying to find functionality without knowing exact variable names

06Self-hosted infrastructure using Docker and Qdrant for data privacy

Use Cases

01Onboarding to a large, legacy codebase to find entry points and patterns

02Locating distributed implementations of specific behaviors across multiple files

03Tracing complex data flows when keyword-based grep returns too many irrelevant results

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add adilasif/local-semantic-search-claude-code-plugin semantic-explore

For use in Claude.ai and ChatGPT

Key Features

010 GitHub stars

02GPU-accelerated semantic search for rapid code indexing and retrieval

03Relevancy scoring to filter strong matches from noise

04Tailscale support for searching remote codebases securely

05Intent-based querying to find functionality without knowing exact variable names

06Self-hosted infrastructure using Docker and Qdrant for data privacy

Use Cases

01Onboarding to a large, legacy codebase to find entry points and patterns

02Locating distributed implementations of specific behaviors across multiple files

03Tracing complex data flows when keyword-based grep returns too many irrelevant results

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add adilasif/local-semantic-search-claude-code-plugin semantic-explore

For use in Claude.ai and ChatGPT