LOCAL RAG PROJECT · OLLAMA OR LLAMA.CPP · CHROMA RETRIEVALLOCAL INFERENCE · URL FETCHING NEEDS NETWORKCodeMind source ↗
Local AI · Open-source Python project

CodeMind: a local RAG learning project

Explore a retrieval-augmented generation workflow that ingests web pages or local text files, retrieves relevant passages, and uses a local model to help answer questions. Local inference does not guarantee correctness; fetching web pages requires network access.

HOW THE REPOSITORY WORKS

CodeMind retrieval pipeline

Local model inference · inspectable source
RAG · INGEST → CHUNK → RETRIEVE → ANSWERChromaDB + Ollama or llama.cpp
01 / SOURCE→

Web page or text file

Local PDFs, source files, and Markdown specs

./docs/**/*
02 / EMBED→

Extract and chunk text

Split text and create local embeddings

nomic-embed-text
03 / STORE→

Local Chroma index

Persist searchable vectors on the host

ChromaDB
04 / INFERENCE→

Local language model

Generate answers using a local Ollama model

localhost:11434
05 / ANSWER✓

Answer with context

Return grounded answers with source references

local output

Local-first workflow

Documents and inference stay on the configured machine; no cloud API is needed for the local path.

Runs on your configured machine

Bring your own model

Connect a supported local model runtime and select the model that fits your hardware.

Runs on your configured machine

Grounded answers

Use retrieved passages as context and inspect the source material behind a response.

Runs on your configured machine

Persistent index

Keep an on-disk vector index for repeat queries against the same workspace.

Runs on your configured machine
HANDS-ON LOCAL WORKFLOW

Install, ingest, and ask

Follow the project's documented CLI flow. The commands below run the interactive Python module and are entered at the CodeMind prompt.

HANDS-ON LOCAL WORKFLOW

Start with three terminal steps.

Keep the project files on your machine, use a local inference runtime, and follow the setup for the version you have installed.

Create an environment, install dependencies, and launch the interactive CLI from the repository root.

terminal · codemind
git clone https://github.com/SilentCoder-HI/codemind.git
cd codemind
python -m venv env
source env/bin/activate
pip install -r requirements.txt
python -m main
MORE OPEN-SOURCE TOOLS

Explore the rest of the toolkit

CodeMind is one of three projects in the SilentCoder ecosystem. Visit the docs or package source for each project.

▣ OPEN SOURCE COLLABORATION

Build Local AI with the Community

Want to contribute to CodeMind or propose an offline developer AI utility? Join our GitHub repository, submit pull requests, or benchmark custom local LLM backends.