similarity-search-patterns Skill for Claude: What It Does and How to Install It

Makes Claude build semantic search with Pinecone, Qdrant, pgvector or Weaviate, and pick the right index.

Installs: 10.9K; on skills.sh
Stars: 42.1K; on GitHub
Files: 1; besides SKILL.md
Updated: Oct 2026; last repo commit

This skill is a vector search guide for Claude, from wshobson's agents collection. SKILL.md compares four distance metrics (cosine, Euclidean, dot product, Manhattan) and three index types (flat, HNSW, IVF+PQ). It ends with do's and don'ts such as using HNSW for most cases, measuring recall and pre-filtering.

references/details.md has four Python templates: Pinecone, Qdrant, pgvector with PostgreSQL, and Weaviate. The Qdrant, pgvector and Weaviate templates include a hybrid search method. It loads when you build semantic search, RAG retrieval or recommendations, or tune retrieval speed.

Good for

Developers building semantic search or RAG retrieval
Teams adding recommendations to a product
Engineers scaling search to millions of vectors
Developers mixing keyword and vector search

Install similarity-search-patterns

Claude Code

Copy this prompt.

Install the similarity-search-patterns skill in Claude Code. Run this command:

/plugin marketplace add wshobson/agents
/plugin install llm-application-dev@claude-code-workflows

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Pick one route. The plugin routes install the whole llm-application-dev plugin: 3 agents, 3 commands and 8 skills (counted from the plugin folder in the repo tree, which has no hooks or MCP config files). The skills-only routes install just this skill, with no agents or commands. The skill needs Python and a vector database of your own. The templates use Pinecone, Qdrant, PostgreSQL with pgvector or Weaviate.

From the project README. Source.

What's in the skill

When it loads

Implement efficient similarity search with vector databases. Use when building semantic search, implementing nearest neighbor queries, or optimizing retrieval performance.

Sections in SKILL.md

When to Use This Skill
Core Concepts
1. Distance Metrics
2. Index Types
Templates and detailed worked examples
Best Practices
Do's
Don'ts

Key facts

Author: wshobson (repo owner)
License: MIT
Installs on skills.sh: 10,873 (Nov 2, 2026)
GitHub stars: 42,081 for the whole wshobson/agents repo (Nov 2, 2026)
Last repo commit: Oct 30, 2026
What it includes: 2 files including SKILL.md: references/details.md (no scripts)
Needs: Python and a vector store of your own. The templates use Pinecone (needs an API key), Qdrant, PostgreSQL with pgvector or Weaviate
Part of: The llm-application-dev plugin (version 2.0.7) in wshobson/agents. Counted from the plugin folder: 3 agents, 3 commands and 8 skills
Works in: Claude Code, Codex CLI, Cursor, OpenCode, Antigravity CLI, GitHub Copilot and Pi, per the README. Skills-only installers ask which tool you use

What it includes

Cosine, Euclidean, dot product and Manhattan, and when each fits
A table that matches each metric to the kind of vectors it suits, such as cosine for normalized embeddings.
Read · Distance metrics
Flat, HNSW and IVF+PQ compared by speed, recall and data size
A diagram that compares exact search with graph-based and quantized indexes.
Try: “Which index should I use for 5 million embeddings?”
Read · Index types
Store with upsert, search, rerank and delete (details.md)
Takes an API key, creates an index in your Pinecone account if it is missing, and can delete vectors by ID or by filter.
Write · Template 1: Pinecone
Store with filtered and sparse-vector search (details.md)
Connects to a Qdrant host and port. Includes a hybrid search with dense and sparse vectors.
Write · Template 2: Qdrant
PostgreSQL store with an HNSW index (details.md)
Runs CREATE EXTENSION, CREATE TABLE and CREATE INDEX statements, then upsert, search and hybrid search with full-text.
Write · Template 3: pgvector
Store with a schema and BM25 plus vector search (details.md)
Connects to a Weaviate URL (http://localhost:8080 by default) and includes a hybrid search method.
Write · Template 4: Weaviate
Do's and don'ts for index choice, tuning and cost
Covers tuning ef_search and nprobe, hybrid search, recall checks, pre-filtering, p99 latency and vector storage cost.
Read · Best practices

From the official docs

Permissions

Low risk

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

Templates create and delete data in your stores
details.md has a Pinecone create_index call, a delete_by_filter method, and pgvector setup statements that create an extension, a table and an index. Run them against a test index or database first. github.com
The Pinecone template takes an API key
The class takes api_key as an argument. Load it from an environment variable or secrets manager instead of writing it into code. github.com
The plugin route installs more than this skill
The llm-application-dev plugin has 3 agents, 3 commands and 8 skills. We found no hooks or MCP config files in its folder. The skills-only routes install only this skill. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	No read-only mode. Use version control.
Talks to	Pinecone, Qdrant, Weaviate or PostgreSQL, whichever your code connects to
Sign-in and keys	SKILL.md does not mention any account. The Pinecone template takes an api_key argument.
Scopes	Claude can write vector search code in your project when you ask The Pinecone template can create an index in your Pinecone account and delete vectors by ID or filter The pgvector template runs CREATE EXTENSION, CREATE TABLE and CREATE INDEX on your PostgreSQL database

Questions

What does the similarity-search-patterns skill do?

It gives Claude patterns for vector search. Claude uses them to pick a distance metric and index, and to write a search layer on Pinecone, Qdrant, pgvector or Weaviate.

How do I install similarity-search-patterns?

Use the install box at the top of this page. Pick a skills-only route to get just this skill, or the plugin route to get the whole llm-application-dev plugin.

Is this skill from Anthropic?

No. It is in wshobson/agents, a community collection. The skill sits in the llm-application-dev plugin there.

Do I need an API key?

Only for Pinecone. The Pinecone template takes an api_key argument. The Qdrant, pgvector and Weaviate templates take a host, a connection string or a URL instead.

How is it different from vector-index-tuning?

This skill covers building the search layer on four vector stores. vector-index-tuning, a sibling in the same plugin, covers tuning HNSW settings, quantization and index monitoring. Neither SKILL.md names the other.

Does it work in the Claude app?

The install box has no Claude app upload for it, because the skill is built around Python code that runs in your own project. Use Claude Code or Codex. The README also lists Cursor, OpenCode, Antigravity CLI, GitHub Copilot and Pi.

Is it safe?

It bundles no scripts: just SKILL.md and one reference file. When you ask for code, Claude writes it in your project. The templates can create a Pinecone index, run PostgreSQL CREATE statements and delete vectors by filter, so review each before you run it (our advice).

Information

updated Nov 2, 2026

Repository: wshobson/agents
Folder: plugins/llm-application-dev/skills/similarity-search-patterns

wshobson/agents skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 4

similarity-search-patterns SKILL.md · github.com
wshobson/agents README · github.com
llm-application-dev plugin.json · github.com
similarity-search-patterns references/details.md · github.com