chroma

Store and query vector embeddings with metadata filtering for semantic search.

1|Updated May 12, 2026

Search Tags:#semantic search #embeddings #rag #vector database #ai memory #chroma

Ready to run

npx skills add https://github.com/projectedanx/hermes-agent --skill chroma-projectedanx

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: chroma
Source: https://github.com/projectedanx/hermes-agent/tree/main/optional-skills/mlops/chroma
Command: npx skills add https://github.com/projectedanx/hermes-agent --skill chroma-projectedanx

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires chromadb, sentence-transformers, and includes references (resource) components.

What problem does it solve?

This skill addresses the challenge of managing long-term memory and semantic retrieval for AI applications by providing a robust, local-first vector database.

Core Features & Use Cases

Vector Storage: Efficiently store and retrieve embeddings with associated metadata.
Semantic Search: Perform high-speed similarity searches to power RAG (Retrieval-Augmented Generation) workflows.
Use Case: Build a local document assistant that can query thousands of internal PDFs or text files by semantic meaning rather than just keyword matching.

Quick Start

Use the chroma skill to initialize a persistent database at the path ./my_data and add a collection for storing your document embeddings.

Frequently Asked Questions about chroma

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I implement semantic search for local documents in my AI application?▼

You can implement semantic search by storing document embeddings and associated metadata in a local vector database, enabling high-speed similarity searches to retrieve information based on semantic meaning rather than keyword matching.

What is the best way to add persistent memory to an LLM workflow?▼

Persistent memory for LLMs is achieved by using a vector database to store embeddings, which facilitates retrieval-augmented generation workflows and allows the model to access long-term contextual information.

Does this vector database approach work with LangChain and LlamaIndex frameworks?▼

Yes, this open-source embedding database supports integration with major frameworks like LangChain and LlamaIndex, allowing you to easily connect your vector storage to existing AI application pipelines.

How do I store and filter vector embeddings with metadata for RAG?▼

You can efficiently store and retrieve embeddings alongside their associated metadata, enabling metadata-filtered retrieval to fetch specific vector data for your retrieval-augmented generation workflows.

Can I run a vector database locally for querying thousands of internal PDFs?▼

Yes, this skill provides a local-first vector database solution, allowing you to initialize a persistent database at a specific path and build a document assistant to query large volumes of internal text files.