World Monitor is an open source, self-hostable situational awareness dashboard that pulls 500+ curated news feeds, market data and geospatial layers into one screen, and today it is the top repository on GitHub trending with roughly 4,100 new stars in a single day on top of 69,700 total. Setup takes about five minutes if you just want the desktop app, or about fifteen if you want the full Dockerised stack running on your own box. This is our daily Tutorials entry: one trending GitHub tool, installed for real.
- Three install paths: a signed desktop binary for Windows, macOS and Linux, a plain npm dev server that needs zero environment variables, and a Docker Compose stack for a permanent self-hosted instance.
- The dev server is genuinely dependency-light:
git clone,npm install,npm run dev, and the dashboard is live on localhost:3000 with no API keys. - Six site variants ship from one codebase (world, tech, finance, commodity, happy, energy), each with its own npm script.
- Licence is AGPL-3.0-only, so commercial and SaaS use is allowed but carries copyleft and source-availability obligations. Read that before you build a product on it.
What is World Monitor and why is it trending?
World Monitor, by Elie Habib, describes itself as a real time global intelligence dashboard. In practice it is a single screen that stitches together news aggregation, a 3D globe and a WebGL flat map, market data and a set of derived risk scores. The repository lists 500+ curated feeds across 15 categories, a dual map engine built on globe.gl with Three.js and deck.gl with MapLibre GL, 56 map layer types, a finance radar covering 29 stock exchanges, and a Country Instability Index that scores 31 Tier-1 countries. It aggregates 65+ external providers behind those feeds.
RelatedStrix Setup: Run an AI Penetration Tester on Your Code
The reason it is climbing right now is less about any single feature and more about the category. Open source OSINT dashboards usually stop at a map with pins on it. This one bundles the AI summarisation layer, the correlation logic across military, economic and disaster streams, an MCP server so agents can query it, and a native desktop build, then puts the whole thing under a copyleft licence. The repo tags itself with palantir, which tells you exactly who it is aimed at.
One detail worth knowing before you commit: it can run its AI layer through Ollama locally, so summarisation does not require an API key. Groq and OpenRouter are supported as hosted alternatives if you want them.
What do you need before you install?
For the desktop app, nothing. For the source install you need Node and npm. For the Docker route the project's own self-hosting guide lists Docker or Podman (rootless is fine), Docker Compose or podman-compose, and Node.js 22+ on the host for the seed scripts. That Node requirement catches people out because they assume the container handles everything; the seeders run outside it.
How do you install the desktop app on Windows, macOS or Linux?
The desktop build is Tauri 2 with a Node.js sidecar, and one binary switches between variants in-app. The latest tagged release is v2.5.23, published 2026-03-01, with eight assets. Pick the one for your platform:
# Windows: World.Monitor_2.5.23_x64-setup.exe (26.9 MB)
# Windows MSI: World.Monitor_2.5.23_x64_en-US.msi (37.9 MB)
# macOS Apple Silicon: World.Monitor_2.5.23_aarch64.dmg (41.7 MB)
# macOS Intel: World.Monitor_2.5.23_x64.dmg (43.4 MB)
# Linux x86_64: World.Monitor_2.5.23_amd64.AppImage (195.7 MB)
# Linux arm64: World.Monitor_2.5.23_aarch64.AppImage (187.0 MB)
The AppImage is an order of magnitude larger than the Windows installer because it bundles its own runtime. Note the release date: the tagged desktop builds sit at 2.5.23 from March, while the default branch's package.json already reads 2.10.0. The web app and the source tree are considerably ahead of the binaries, which matters if you are chasing a feature you saw in a screenshot.
How do you run it from source?
This is the path we would recommend for a first look, because it is four commands and the project explicitly states the app runs with no environment variables:
git clone https://github.com/koala73/worldmonitor.git
cd worldmonitor
npm install
npm run dev
Then open http://localhost:3000. If port 3000 is taken, set DEV_PORT in .env.local; it is deliberately not VITE_-prefixed so it never leaks into the client bundle. Feature-specific data sources want credentials, and .env.example lists them all with the free-tier signup link next to each one, but every key is optional.
To run one of the sibling variants instead of the full world view:
npm run dev:tech # tech.worldmonitor.app
npm run dev:finance # finance.worldmonitor.app
npm run dev:commodity # commodity.worldmonitor.app
npm run dev:happy # happy.worldmonitor.app
npm run dev:energy # energy.worldmonitor.app
How do you self-host the full stack with Docker?
The dev server is a front end. The full stack adds Redis and the AIS relay, and the project's SELF_HOSTING.md is blunt that three secrets are mandatory or containers will exit on boot:
git clone https://github.com/koala73/worldmonitor.git
cd worldmonitor
npm install
# generate the REQUIRED secrets
echo "RELAY_SHARED_SECRET=$(openssl rand -hex 32)" >> .env
echo "REDIS_PASSWORD=$(openssl rand -hex 32)" >> .env
echo "REDIS_TOKEN=$(openssl rand -hex 32)" >> .env
docker compose up -d # or: uvx podman-compose up -d
./scripts/run-seeders.sh
Then open http://localhost:3000 again. API keys go in a docker-compose.override.yml, which is gitignored, so your secrets stay off the repo. There is also a programmatic surface if you want the data without the UI:
npx worldmonitor tools # list every MCP tool, no key needed
npm install -g worldmonitor # installs the wm command
worldmonitor risk IR --api-key wm_xxx
How does it compare to the alternatives?
| World Monitor | Liveuamap | News aggregators | Commercial OSINT suites | |
|---|---|---|---|---|
| Source available | Yes, AGPL-3.0 | No | Varies | No |
| Self-hostable | Yes, Docker or source | No | Some readers | Enterprise deployment only |
| Geospatial layers | 3D globe plus flat map, 56 layer types | Map first | None | Yes |
| Cross-stream correlation | Yes, with derived risk index | No | No | Yes |
| Local AI option | Yes, via Ollama | No | Rare | No |
| Agent and API access | MCP server, REST, four SDKs | No | RSS only | Contract dependent |
| Desktop binary | Yes, Tauri 2 | No | Some | Varies |
The honest framing is that this is not competing with a news reader. It is an open, self-hostable approximation of the expensive category, with the tradeoffs that implies: you own the deployment, the data quality depends on feeds you can inspect, and nobody is on call when a provider changes its API.
RelatedVibe-Trading Setup: Self-Host an AI Trading Agent Free
What are the gotchas before you rely on it?
The licence is AGPL-3.0-only, not MIT. Self-hosting, forking and even commercial use are all permitted, but the copyleft and source-availability obligations travel with it. If you put a modified instance behind a network service, AGPL section 13 is the clause to read. The project offers separate commercial licensing for teams that need non-AGPL terms.
A previously documented default token was removed. Earlier releases shipped wm-local-token as the default REST token. That default is gone as of PR #3804 because anyone who rebound the proxy from 127.0.0.1 to 0.0.0.0 was authenticating strangers with a publicly documented string. Fresh clones and existing ones both need real values in .env.
Do not disable relay auth on a reachable host. There is an I_UNDERSTAND_THIS_DISABLES_AUTH=true escape hatch for local debugging. It logs a loud security warning every five minutes for a reason.
The binaries lag the source. Tagged releases stop at v2.5.23 from March while the branch is at 2.10.0. If you want current behaviour, run from source.
Feed volume is the real cost. 500+ feeds across 65+ providers is a lot of outbound requests. Redis caching is in the stack for a reason, and the project ships a freshness monitor across 35 source groups because sources do go stale.
- Release cadence for binaries. A five month gap between the tagged desktop build and the branch is the clearest thing to fix. Watch whether v2.10 ships as a binary.
- The MCP surface. Public
tools/listwith authenticatedtools/callis a sensible split. If agent traffic becomes the main consumer, expect the pricing page to matter more than the dashboard. - Whether the risk scoring gets audited. A Country Instability Index that scores 31 countries is a strong claim. Independent scrutiny of the methodology is what would move it from impressive to trustworthy.
Our take
Most trending dashboards are screenshots with a repo attached. This one installs cleanly, runs without a single API key, and ships a documented Docker path with mandatory secrets rather than convenient defaults, which is a better security posture than most projects at this star count manage. The four-command dev install genuinely works and is the right way to evaluate it.
Where we would slow down is the derived intelligence. Aggregating 500 feeds is engineering; scoring national instability is a claim, and a number on a dark dashboard reads as more authoritative than its inputs justify. Treat the map, the feeds and the market data as the product, and treat the composite scores as a hypothesis until the methodology has been picked apart in public. Run it from source, keep the secrets real, and read the AGPL before it ends up inside something you sell.
- Repokoala73/worldmonitor on GitHub README, quick start, tech stack and licence table
- ReleasesLatest release, v2.5.23 desktop binary names and file sizes
- OfficialWorld Monitor self-hosting guide deployment options beyond the dev server
- Referenceworldmonitor on npm the CLI and MCP tool listing
- ReferenceTauri 2 the framework behind the desktop builds
Original analysis by GenZTech. Commands and file sizes verified against the project's README, SELF_HOSTING.md and GitHub Releases.
