Skip to content

Shinkai is a two click install AI manager (Ollama compatible for Windows, Mac and Linux). It lets you download/use AI models, RAG, and performs actions for you with tooling (very soon).

License

Notifications You must be signed in to change notification settings

dcSpark/shinkai-node

Repository files navigation


Shinkai Node

The Shinkai Node is the central unit within the Shinkai Network that links user devices and oversees AI agents. Its diverse functions include processing user inputs, managing AI models, handling external containerized tooling for AI, coordinating computing tasks, generating proofs, converting and indexing data into vector embeddings, and ensuring efficient task execution according to user needs. The nodes play a crucial role in maintaining the network's decentralized structure, which enhances both security and privacy.

There is a companion repo called Shinkai Apps, that allows you to locally run the node and also easily manage AI models using Ollama, you can find it here.


Requirements

Rust

The Shinkai Node requires Rust 1.76.0 or later.

GCC Compiler Setup

Make sure you have gcc/g++ as your default compilers:

export CC=gcc
export CXX=g++

Build

Easy Build

sh scripts/run_node_localhost.sh

if you want to restart the node, you can delete the folder storage and run the build again. More information at https://docs.shinkai.com/getting-started.

Build Shinkai Rust Node

cargo build

Note: You must run this command from the root directory of this repo and make sure that you have set the required ENV variables.

Add Telemetry support

cargo build --features telemetry

Tests

Note: You must run these tests from the root directory of this repo.

Testing All Sub-projects Locally

Simply run the following command to run tests for all projects:

sh scripts/test_all_subprojects.sh

Test Shinkai Rust Node Only

Simply use the following to run all rust node tests:

cargo test -- --test-threads=1

For running a specific test (useful for debugging) you can use:

cargo test tcp_node_test -- --nocapture --test-threads=1

Running Dockerized Tests

Main tests

# Build testing image
docker build -t testing_image -f .github/Dockerfile .

# Run tests main cargo tests
docker run --entrypoint /entrypoints/run-main-cargo-tests.sh testing_image

WASM tests

# Build testing image - shinkai-message-wasm
docker build -t testing_image_wasm -f .github/Dockerfile.wasm .

# Run tests shinkai-message-wasm cargo tests
docker run --entrypoint /entrypoints/run-wasm-pack-tests.sh testing_image_wasm

# Run tests shinkai-message-wasm wasm-pack tests
docker run --entrypoint /entrypoints/run-wasm-cargo-tests.sh testing_image_wasm

Shinkai App tests

You need to compile the wasm library from shinkai-message-wasm and copy the resulting pkg to shinkai-app/src/pkg (if a folder already exists you should delete first). Then you will be able to run the tests inside shinkai-app

npm run test.unit

Shinkai PYO3 Tests

Further CI Development

Use act -j test-wasm -P self-hosted=nektos/act-environments-ubuntu:18.04 --container-architecture linux/amd64 to run the tests locally in a docker container. This is useful for debugging CI issues.

About

Shinkai is a two click install AI manager (Ollama compatible for Windows, Mac and Linux). It lets you download/use AI models, RAG, and performs actions for you with tooling (very soon).

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published