Platform

Platform is shared infrastructure for student orgs. One deployment serves many orgs. Each org is a Discord server and turns on only the modules it uses.

Each org route checks access. The audit log records each change that an officer or a token makes. Org secrets are encrypted in the database.

Modules

Module	Does
points, users	Members, event points, leaderboards, CSV imports
storefront	A merch store paid in points, with prices checked on the server
calendar	Notion events synced to Google Calendar, with the credentials of each org
leetcode, games	The daily LeetCode post and Jeopardy in Discord
organizations, superadmin	Orgs, officers, module switches, secrets and machine tokens; for the superadmin, the server's services and the shared apps
godfather	GPU and CPU pods on the org's RunPod account. Members connect over SSH with 12-hour certificates
agents	Conversations, memories and a profile graph for each member, for agents that talk to members
knowledge	Hybrid search (pgvector and full text) over documents and crawled public pages
submodules	Shared content from submodules/: content made once for every org, such as the ASU campus pages and live queries for knowledge, and the job and hackathon feeds for Feeds
alerts	Org events, such as a failed job or a site that is down, sent to the org's Discord destinations
accounts	Canvas, Google and Outlook sign-in for a member, so that agents can act for them
feeds	New job listings and hackathons posted to an alert destination
runpod	Hosting: deploys of an org's apps to RunPod or This server (capped containers on the deployment's server), with health checks and rollback; shared apps such as the vault
uptime	Checks of sites and Hosting apps on a schedule, with events when one goes down or up
dashboard	One page for each org with problems, activity, jobs, CI runs and module state
mcp	An MCP server and /api/tools that give agents the module tools through scoped machine tokens
auth, public, bot	Discord sign-in, tokens and access checks; open reads for public pages; the Discord bot

The dashboard calls modules features and submodules shared content. Each module is in one category: Core, Storage, AI and agents, Automations, Bots or Compute. Core modules are always on. A new org starts with each other module off. A module can read submodules from submodules/: content made once for every org. See Architecture.

Processes

Process	Entry point	Port
API	main.py (gunicorn)	8000
Discord bot	bot_main.py
Job worker	worker_main.py (Postgres only)
MCP server	mcp_main.py	8001
Dashboard and member store	dashboard/	5001 (5173 in dev)

The database is Postgres, or SQLite for a small deployment. Alembic makes the schema.

Start

You need Podman with podman-compose (or Docker), make and uv.

git clone https://github.com/asusoda/platform.git
cd platform
uv sync
uv run pre-commit install
cp .env.template .env      # Discord app, bot token, secrets
make dev

The API is at http://localhost:8000 and the dashboard at http://localhost:5001. Create an org with flask --app main org create. Getting started has the settings and the steps to run Platform on one RunPod pod.

Commands

make dev       # start with logs
make up        # start in the background
make down      # stop
make check     # fix lint and format, then type check, tests, migrations
make ci        # the checks that CI runs, with no changes to files
make shell     # shell in the API container
make deploy    # deploy on the server

Documentation

docs/ has the guides. Each module folder has a README.md. docs/roadmap.md has the plan.

Deployments

The Software Developers Association (SoDA) at ASU and AI Society at ASU run Platform. Their servers are the examples in these docs.

License

BSD 3-Clause (modified for web attribution). Copyright The Software Developers Association at ASU. See LICENSE.