Skip to main content

Self-hosted AI orchestration platform — agentic runtime you own end-to-end

Project description

Etherion

Etherion

The self-hosted agentic platform. You own it entirely.

PyPI etherion PyPI etherion-tui


Etherion is an autonomous, goal-oriented AI platform you deploy on your own hardware. Give it a business objective — it decomposes, executes, and synthesises the result using a team of AI specialists. No task breakdowns. No SaaS lock-in. No data leaving your infrastructure.

User: "Analyse my top 50 customers for churn risk and draft personalised emails"

Platform: → Decomposes goal → Selects specialists → Executes in parallel
          → Reads your CRM (Jira / HubSpot / Shopify) → Writes emails
          → Delivers results. Done.

Install

pip install etherion etherion-tui

Quick Start

# 1. Scaffold config
etherion init

# 2. Edit .env (API keys, database password, etc.)

# 3. Boot infrastructure  (Postgres + Redis + MinIO via Docker Compose)
etherion bootstrap --mode docker

# 4. Run migrations
etherion migrate

# 5. Create first tenant
etherion tenant create --email you@example.com --password yourpass

# 6. Start API
etherion serve

# 7. Open the terminal UI
etherion-tui

That's it. Visit http://localhost:8080/graphql for the API playground.


Architecture

┌──────────────────────────────────────────────────────────────────┐
│  Terminal UI (etherion-tui)  ·  Frontend (Next.js / LobeChat)   │
└────────────────────────┬─────────────────────────────────────────┘
                         │ GraphQL + REST + WebSocket
┌────────────────────────▼─────────────────────────────────────────┐
│                    FastAPI  ·  Strawberry GraphQL                 │
│            JWT auth  ·  CSRF  ·  Rate limiting  ·  RLS           │
└──┬──────────┬──────────┬──────────┬──────────────────────────────┘
   │          │          │          │
   ▼          ▼          ▼          ▼
PostgreSQL  Redis     MinIO       Vault
+ pgvector  Pub/Sub   Objects     Secrets
  RLS       Celery    Artifacts

Every tenant is fully isolated at the database layer via Row-Level Security — not just application logic. A bug in application code cannot cause cross-tenant data leaks.


What Makes It Different

Goal-Oriented Execution

The Orchestrator agent receives a goal, builds a checklist, picks specialists, and executes — no manual task decomposition required.

Natural Language Agent Creation

"Create an agent that watches my Shopify store for abandoned carts
 and sends personalised recovery emails with a 10% discount code."

The platform designs the architecture, connects your tools, wires the knowledge base, and deploys it.

Knowledge Base Without Hallucinations

Every query forces a live web search alongside retrieval. The platform never answers from stale embeddings alone.

Full Operator Control

  • Run on bare metal, VMs, or Docker
  • All data stays on your infrastructure
  • Secrets injected into RAM via Vault — never written to disk
  • OAuth apps registered by YOU, not by us

Features

Area Details
Orchestration Checklist-based 2N+1 loop, parallel specialist dispatch, full audit trail
Knowledge base pgvector semantic search, per-tenant RLS, multimodal (PDF, images, text)
OAuth integrations GitHub, Google, Slack, Microsoft 365, Notion, Jira, HubSpot, Linear, Shopify
Object storage MinIO — per-tenant buckets, presigned URLs, artifact management
Secrets HashiCorp Vault — secrets injected at runtime, never persisted in env
TUI Bubble Tea — onboarding wizard, OAuth manager, live job monitor, tenant switcher
CLI etherion init / bootstrap / migrate / serve / tenant / status
API FastAPI + GraphQL subscriptions, REST, WebSocket, multipart upload
Auth JWT, multi-provider OAuth login, invite-only onboarding, per-tenant RLS
Real-time Redis Pub/Sub → GraphQL subscriptions → live execution traces in UI

CLI Reference

etherion init          Scaffold .env and docker-compose in current directory
etherion bootstrap     Start infrastructure (--mode docker | nix)
etherion migrate       Run Alembic database migrations
etherion serve         Start the API server
etherion worker        Start Celery worker  (start | beat)
etherion tenant        Manage tenants and users
etherion status        Health-check: Postgres, Redis, MinIO, API, Worker
etherion where         Print config and binary paths

Deployment Modes

Mode Command When to use
Docker --mode docker First install, dev, single server
Nix --mode nix Bare-metal, fully declarative, reproducible prod

For production, the Nix mode provisions the full stack declaratively: NixOS → Matchbox → Ansible → systemd + MinIO + Vault.


Documentation

  • SETUP.md — Local and production setup guide
  • Z/agents.md — Agent-first contribution workflow (read before contributing)
  • Z/tech.md — Complete technical architecture
  • Docs/etherion_docs/ — Full platform documentation (12 sections)

Contributing

Etherion uses an agent-first contribution model. Before writing a single line, your AI agent must read:

  1. Z/tech.md — full architecture
  2. Z/agents.md — contribution workflow and standards

Every PR must include a contribution log (Logs/Daily/<your-email>) and a Z/tech.md update. See Z/agents.md for the full workflow.


License

MIT


Contact

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

etherion-0.1.6-py3-none-any.whl (1.4 MB view details)

Uploaded Python 3

File details

Details for the file etherion-0.1.6-py3-none-any.whl.

File metadata

  • Download URL: etherion-0.1.6-py3-none-any.whl
  • Upload date:
  • Size: 1.4 MB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.14.3

File hashes

Hashes for etherion-0.1.6-py3-none-any.whl
Algorithm Hash digest
SHA256 8020fb6ea6de45a0047759b60aaddd7b25452d504d629aa566e0075cc191cc2e
MD5 5c5b9a955f110ce0152a74085dfcb356
BLAKE2b-256 6cf9bd1d71723bdf3b468a725c77ecfc4e47695da645fa4cd89e7ec6adb3a5e3

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page