Install

Both packages, and which extras you actually need.

Two packages. Agentino is the agent framework; Runspace is the workspace that runs agents and gives them a UI. You can use either on its own.

Agentino

pip install agentino-framework

The distribution is agentino-framework because the agentino name on PyPI belongs to an unrelated project. It imports as agentino regardless.

Runspace

pip install "runspace[agentino,workspace,server]"

Which extras you need — Agentino

The core install is three dependencies. Everything heavier is opt-in, and agentino.tools.std imports each lazily, so a tool you do not use costs nothing.

Extra	Pulls in	Use it when
`docgen`	reportlab, openpyxl, python-docx, python-pptx, markdown	Agents produce PDFs, spreadsheets, documents or slides
`memory`	numpy	The knowledge base uses dense embeddings
`web`	feedparser, beautifulsoup4	Web search, RSS and page fetching
`telegram`	aiogram	The Telegram transport
`slack`	slack-bolt, aiohttp	The Slack transport, in Socket Mode
`websocket`	aiohttp	The WebSocket transport
`serve`	fastapi, uvicorn, starlette	`--serve`, the webhook and WhatsApp transports
`stdtools`	memory, docgen, web	Every built-in tool
`transports`	telegram, slack, websocket, serve	Every channel
`all`	everything above	You are exploring
`dev`	pytest, pytest-asyncio, ruff	Working on agentino itself

Which extras you need — Runspace

Runspace's core install is deliberately small — pydantic, httpx, pyyaml and a cron parser. Everything else is opt-in:

Extra	Pulls in	Use it when
`agentino`	Agentino	You want agents running in-process rather than shelling out to a CLI
`workspace`	The gateway, registry and messaging	You want the workspace backend, not just the protocol layer
`server`	`uvicorn`, `fastapi`	You are serving over HTTP
`documents`	`reportlab`, `openpyxl`, `python-docx`	Agents produce PDFs or Office files
`redis`	`redis`	Sessions shared across processes
`supabase`	`supabase`	Messages and files persist to Postgres instead of SQLite
`scheduler`	`croniter`	Scheduled routines
`crawler`	`beautifulsoup4`	The web-crawl helpers
`embeddings`	`openai`	`EMBEDDINGS_BACKEND=openai`; without it that backend raises at call time
`all`	everything above	You are exploring
`dev`	the extras the suite exercises, plus agentino, pytest and ruff	Working on runspace itself — a fresh `pip install -e ".[dev]"` gives a checkout where the whole suite runs

Nothing else to configure

Both packages run with no environment set. Runspace falls back to a local SQLite file for messages and a directory on disk for files, so a fresh clone starts and persists without a database. Point SUPABASE_URL and SUPABASE_KEY at a project and the same code writes there instead.

You will need one thing to get an actual answer out of an agent — a model endpoint. The two packages read different variable names, because either can be used without the other:

# Agentino
export AGENTINO_BASE_URL="https://api.openai.com/v1"
export AGENTINO_API_KEY="sk-..."

# Runspace
export AI_BASE_URL="https://api.openai.com/v1"
export AI_API_KEY="sk-..."

Set both if you are using both. Any OpenAI-compatible endpoint works.

Agentino also reads OPENAI_API_KEY as a key fallback, and ~/.agentino/auth.json after that. Its default endpoint is https://api.openai.com/v1 — so if you set only the Runspace names and call Agentino directly, it will quietly use that default rather than your endpoint.