remex-cli 1.2.0

Local-first RAG library — ingest files and SQLite, query semantically, pipe results into any AI agent

remex

Your private knowledge base — fully offline, never leaves your machine.

Semantic search	AI answer
File ingestion	Collections manager

---

Remex is a local-first knowledge base for your documents. Point it at any folder — PDFs, notes, code, spreadsheets — and it becomes instantly searchable using natural language. Ask questions and get answers backed by the exact sources in your files, with no data ever leaving your machine.

It runs entirely offline, requires no cloud account, and works with any AI provider you already have — Anthropic, OpenAI, or a local Ollama instance.

---

Remex Studio

A native desktop app (Windows · macOS · Linux) to ingest, search, and query your documents with AI — no terminal required.

Download the latest release →

Building from source: see studio/README.md.

---

Python CLI

pip install remex-cli          # core — ingest + query
pip install remex-cli[api]     # adds the FastAPI sidecar (used by Studio)

Quick start

# 1. Scaffold a project
remex init

# 2. Drop files into docs/ then ingest
remex ingest docs/

# 3. Semantic search
remex query "how does authentication work?"

# 4. AI answer (auto-detects Anthropic / OpenAI / Ollama)
remex query "how does authentication work?" --ai

All commands

Command	Description
remex init [path]	Scaffold docs/, remex.toml, and .gitignore
remex ingest [dir]	Ingest files from a directory
remex ingest-sqlite <db>	Ingest rows from a SQLite table
remex query <text>	Semantic search (add --ai for AI answer)
remex sources	List all ingested source paths
remex stats	Show chunk/source counts for a collection
remex delete-source <path>	Remove all chunks for a source
remex purge	Remove chunks whose source file no longer exists
remex reset	Wipe an entire collection
remex list-collections	List all collections in a database
remex serve	Start the FastAPI sidecar (used by Studio)

Use remex <command> --help for full option reference.

---

Features

Fully offline	No data leaves your machine — local embeddings, local storage
12 file formats	.pdf .docx .md .txt .csv .json .html .pptx .xlsx .epub .odt .jsonl
SQLite ingest	Embed rows from any table — or all tables at once — alongside your files
Incremental ingest	SHA-256 hash check — unchanged files are skipped automatically
Batch embedding	Chunks are embedded in batches for fast ingestion of large directories
AI answers	Auto-detects Anthropic, OpenAI, or a local Ollama instance
Multi-collection search	Query across collections, results merged by relevance score
Source filtering	Narrow query results to a specific file or table
Export results	Copy or export query results for use elsewhere
Collections manager	Rename, delete, and inspect your knowledge bases from the UI
Keyboard shortcuts	Full keyboard navigation — press ? in Studio for the reference

---

Configuration

Place a remex.toml in your project root (created by remex init):

[remex]
db             = "./remex_db"
collection     = "remex"
embedding_model = "all-MiniLM-L6-v2"

# chunk_size     = 1000
# overlap        = 200
# min_chunk_size = 50
# chunking       = "word"   # "word" or "sentence"

CLI flags always override remex.toml values.

---

_{Changelog · Contributing · Apache 2.0 · GitHub}