Search is all you need.

CosmaSense is a local-first, AI-powered search engine that indexes your files and lets you find information semantically.

CosmaSense macOS app interface showing semantic file search with AI-powered results

See It in Action

Real-world examples of semantic search finding exactly what you need.

Resume Search

You've applied to hundreds of jobs and you're sick of digging through 100+ resume files. Now you're applying to TikTok—just search in CosmaSense, drag the right resume, and you're done. (Pure semantic vector search with natural language—no complicated filters.)

Demo showing semantic search finding resumes by skills and qualifications

Image Search

You're a photographer and you want to upload to Unsplash right now. Instead of hunting folders, just search what's in the image ("golden hour street portrait", "snowy mountain drone shot") and pick the match. (Shows multi-format + media support.)

Demo showing semantic search finding images by content description

Email Attachment Search

Someone sent you a contract last month but you can't remember when or who. Just describe what you're looking for and CosmaSense finds the attachment instantly. (Works across all your downloaded email attachments.)

Demo showing semantic search finding email attachments by content

…and a lot more!

What is CosmaSense?

CosmaSense is an open-source tool that creates a powerful, local index of your files. It uses advanced AI models to understand the content and context, allowing you to search based on meaning, not just keywords. Your data stays private, on your machine.

Why Use CosmaSense?

Unlock the full potential of your local data with features designed for power users and developers.

auto_awesome

Powerful Semantic Search

Go beyond keyword matching. Find files and documents based on concepts, ideas, and context.

lock

Privacy First, Always

Your files are indexed and stored locally. No data ever leaves your machine, ensuring complete privacy.

code

Developer Friendly

Integrate with your workflow through a powerful CLI, TUI, and extensible architecture. It's built by developers, for developers.

Architecture Overview

A modular system designed for flexibility and performance, from the core backend to the native macOS interface.

CosmaSense architecture diagram showing the processing pipeline (Discovery, Extraction, Summarization, Embeddings, Indexing, Hybrid Search), backend core components (File Watcher, AI Engine, Search Engine, REST API), and client interfaces (CLI, TUI, macOS App)

How It Works

From file discovery to intelligent search, here’s a look at the pipeline that powers CosmaSense.

1

Discovery

Continuously monitors specified directories for new or modified files.

2

Parsing

Extracts text and metadata from a wide range of file types, including PDFs, documents, and code.

3

Summarization & Embedding

Local AI models process the extracted content, creating summaries and vector embeddings that capture semantic meaning.

4

Hybrid Search Engine

Combines traditional keyword search with modern vector search to deliver highly relevant results.

Open Source + Community

CosmaSense is built in the open. Join our community to ask questions, contribute code, and shape the future of local-first AI.

Ready to Get Started?

Download the alpha version from GitHub and start exploring your own data with the power of AI.