project-nova
A multi-agent AI architecture that connects 25+ specialized agents through n8n and MCP servers. Project NOVA routes requests to domain-specific experts, enabling control of applications from knowledge bases to DAWs, home automation to development tools. Includes system prompts, Dockerfiles, and workflows for a complete AI assistant ecosystem.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Check it deeper on your machine
It has changed since its scan on 11 Nov 2026.
As a Claude Code plugin, install it and then ask Claude Code:
/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/dujonwalker/project-nova with scanskillsafety
Or from a terminal, clone it and run it:
git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/dujonwalker/project-nova'
It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.
Badge for your README
Markdown
[](/listings/dujonwalker--project-nova)HTML
<a href="/listings/dujonwalker--project-nova"><img src="/api/listings/dujonwalker--project-nova/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Prerequisites
Before setting up Project NOVA, ensure you have:
Option A: OpenWebUI + n8n Setup (Recommended)
- **OpenWebUI Instance**: For the modern chat interface with conversation management - **Note**: For text-to-speech features (voice input), OpenWebUI requires SSL/HTTPS (even a self-signed certificate works) - **n8n Instance**: A running n8n installation (v1.88.0 or later) - **Critical**: Version 1.88.0+ is required as it introduced the native MCP Client Tool node - **No Community Nodes Required**: All agents now use n8n's native MCP Client Tool node
Option B: n8n Only Setup (Basic)
- **n8n Instance**: A running n8n installation (v1.88.0 or later) with chat trigger capabilities - **No Community Nodes Required**: All agents use n8n's native MCP Client Tool node
Common Requirements
- **Docker Host**: Environment for running containerized MCP servers - **LLM Access**: Either: - Cloud API access (OpenAI, Claude, Gemini, etc.) - Local Ollama instance with models that support tool use/function calling - **Applications to Control**: - The actual applications you want NOVA to control (most agents are designed for self-hosted applications) - Examples: Home Assistant, Gitea, Reaper, OBS Studio, TriliumNext, etc. - Each agent requires its corresponding application to be accessible
Setup Instructions
1. Clone the Repository
git clone https://github.com/dujonwalker/project-nova.git cd project-nova
2. Choose Your Deployment Option
**Option A: OpenWebUI + n8n (Recommended)**
This setup provides the best user experience with conversation history, session management, and a modern interface.
1. **Set up OpenWebUI inlet filter**: - In your OpenWebUI instance, navigate to **Workspace** → **Functions** - Click **"Create New Function"** - Copy and paste the code from `openwebui-function/n8n_inlet_filter.py` into the function editor - **Important**: Update the `n8n_url` configuration variable at the top of the filter with your complete n8n webhook URL (including server IP/hostname and webhook UUID) - Save the filter
2. **Create a new model using the filter**: - Navigate to **Workspace** → **Models** - Click **"Create New Model"** - Configure your base model (e.g., your preferred LLM) - In the **Filters** section, select the n8n inlet filter you just created - Save the model - The filter will now automatically forward messages to n8n and manage conversation history
3. **Configure n8n workflows**: - Import the router agent workflow (`n8n-workflows/router_agent.json`) - Set up a webhook trigger in the router agent (configure URL in OpenWebUI filter) - Import specialized agent workflows as needed
**Option B: n8n Only Setup**
1. **Import workflows in your n8n instance**: - **Bulk import via n8n CLI** (easiest): ```bash # Navigate to your n8n installation directory n8n import:workflow --input=path/to/project-nova/n8n-workflows/* ``` - **Import via n8n Web Interface**: - Navigate to your n8n dashboard - Click on "Workflows" in the sidebar - Use the "Import from file" option for each workflow - Start with the router agent first, then specialized agents
2. **Configure after import**: - **Update OpenWebUI filter**: Copy the webhook URL from your imported router agent and paste it into the OpenWebUI inlet filter configuration - **Configure API keys**: Add your actual API keys for the services you want to use (Blinko, Home Assistant, etc.) - **Update server IPs**: Replace `YOUR_SERVER_IP` placeholders with your actual server addresses
3. Set up MCP Servers
- Use the Dockerfiles and docker-compose.yml files provided in `mcp-server-dockerfiles` to build and run your servers - Or use the instructions to set up MCP servers directly on your host - Configure each MCP server for the agents you plan to use
4. Configure Your API Keys and Credentials
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 101 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.
- # Project NOVA
- Project NOVA is a DIY, agent-agnostic toolkit of self-hosted
- [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) servers.
- Each server exposes one focused capability over Streamable HTTP so it can be
- used by the client, harness, or automation runtime that fits your workflow.
- NOVA is **not** a standalone assistant, chat application, central router, or
- required orchestration runtime. Use the servers with Hermes-Agent, llama.cpp
- WebUI, n8n, or any MCP client.
- Optional client-facing skill bundles are indexed in
- [`mcp-skills/README.md`](mcp-skills/README.md).
- ## What changed in v2
- - **25-server MCP catalog:** The repository provides 25 focused MCP servers.
- - **Streamable HTTP baseline:** Servers use Streamable HTTP at `/mcp` where
- supported; each server README documents service-specific transport and
- deployment caveats.
- - **Current implementations:** Each server directory contains its deployment
- files, source or image configuration, and service-specific documentation.
- - **Repository-local skills added:** v2 adds 25 optional skill bundles, one for
- each included MCP server. Use them directly from this repository or copy
- selected bundles into a client-supported local skills directory.
- - **Agent-agnostic usage:** Servers and skills work with compatible clients,
- harnesses, or direct MCP connections.
- ## What is included
- Every active server lives in its own directory under
- [`mcp-server-dockerfiles/`](mcp-server-dockerfiles/). Server READMEs contain
- the service-specific requirements, environment variables, and startup
- instructions. The current transport baseline is Streamable HTTP at `/mcp`;
- consult each server README for its supported transport and endpoint details.
- ### Included server catalog
- | Capability | Directory | Focus |
- | --- | --- | --- |
- | BookStack | [`bookstack-mcp`](mcp-server-dockerfiles/bookstack-mcp/) | Documentation and wiki content |
- | Gitea | [`gitea-mcp`](mcp-server-dockerfiles/gitea-mcp/) | Git repositories, issues, and releases |
- | Forgejo | [`forgejo-mcp`](mcp-server-dockerfiles/forgejo-mcp/) | Forgejo repositories and issues |
- | Karakeep | [`karakeep-mcp`](mcp-server-dockerfiles/karakeep-mcp/) | Bookmarks, tags, and saved content |
- | Memos | [`memos-mcp`](mcp-server-dockerfiles/memos-mcp/) | Notes and lightweight publishing |
- | Paperless-NGX | [`paperless-mcp`](mcp-server-dockerfiles/paperless-mcp/) | Documents, tags, and correspondents |
- | Prometheus | [`prometheus-mcp`](mcp-server-dockerfiles/prometheus-mcp/) | Metrics queries and monitoring |
- | TriliumNext | [`triliumnext-mcp`](mcp-server-dockerfiles/triliumnext-mcp/) | Hierarchical notes and knowledge |
- | Outline | [`outline-mcp`](mcp-server-dockerfiles/outline-mcp/) | Team knowledge-base documents |
- | Home Assistant | [`home-assistant-mcp`](mcp-server-dockerfiles/home-assistant-mcp/) | Smart-home entities and automations |
- | Langfuse | [`langfuse-mcp`](mcp-server-dockerfiles/langfuse-mcp/) | Prompts and observability data |
- | YouTube | [`youtube-mcp`](mcp-server-dockerfiles/youtube-mcp/) | Video metadata and transcripts |
- | Blinko | [`custom-blinko-mcp`](mcp-server-dockerfiles/custom-blinko-mcp/) | Notes, flash notes, and sharing |
- | Fetcher | [`fetch-mcp`](mcp-server-dockerfiles/fetch-mcp/) | Web content retrieval |
- | Playwright | [`playwright-mcp`](mcp-server-dockerfiles/playwright-mcp/) | Browser automation |
- | Filesystem | [`filesystem-mcp`](mcp-server-dockerfiles/filesystem-mcp/) | File and directory operations |
- | OBS | [`obs-mcp`](mcp-server-dockerfiles/obs-mcp/) | Scenes, sources, and recording |
- | Ableton Live | [`ableton-copilot-mcp`](mcp-server-dockerfiles/ableton-copilot-mcp/) | Music production control |
- | REAPER | [`reaper-mcp`](mcp-server-dockerfiles/reaper-mcp/) | Live REAPER project control |
- | REAPER Project QA | [`reaper-qa-mcp`](mcp-server-dockerfiles/reaper-qa-mcp/) | Offline REAPER project analysis |
- | CLI/Shell | [`cli-server-mcp`](mcp-server-dockerfiles/cli-server-mcp/) | Controlled command-line execution |
- | Nextcloud | [`nextcloud-mcp`](mcp-server-dockerfiles/nextcloud-mcp/) | Files, calendars, and collaboration |
- | Komodo | [`komodo-mcp`](mcp-server-dockerfiles/komodo-mcp/) | Infrastructure and deployment management |
- | SearXNG | [`searxng-mcp`](mcp-server-dockerfiles/searxng-mcp/) | Federated web search |
- | Vision Bridge | [`vision-bridge-mcp`](mcp-server-dockerfiles/vision-bridge-mcp/) | Image understanding and analysis |
- ## Quick start
- 1. Choose a server from the catalog and read its server-specific README.
- 2. Copy or create the required environment configuration for the service.
- 3. Start the server with its Docker Compose configuration:
- ```bash
- cd mcp-server-dockerfiles/<server-directory>
- docker compose up --build
- ```
- 4. Register the server's `http://<host>:<port>/mcp` endpoint in your MCP
- client or compatible harness.
- 5. Test the connection with a read-only operation before enabling writes or
- destructive actions.
- Most servers require the application they control to be running separately.
- Credentials and service URLs are intentionally configured per server rather
- than through a central NOVA runtime. Copy the relevant `.env.example` to
- `.env` locally; `.env` files are ignored and must never be published.
- ## Repository layout
- ```text
- project-nova/
- ├── README.md
- ├── mcp-skills/ # Optional client-facing skill bundles
- │ └── <skill-name>/SKILL.md
- └── mcp-server-dockerfiles/ # Active MCP server implementations
- └── <server>-mcp/
- ├── Dockerfile
- ├── docker-compose.yml
- ├── README.md
- └── startup/source files
- ```
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.