galagos-cli 0.1.0a3

Command-line client for the Galagos AI bioinformatics platform.

galagos-cli

Command-line client for the Galagos AI bioinformatics platform. Drives sandboxes, chats, and files programmatically against the Django REST API.

Install

pipx install galagos-cli      # recommended
# or
pip install galagos-cli

For local development:

python3.11 -m venv .venv
.venv/bin/pip install -e .

Quick start

# 1. Point at the API and paste a token (mint one from the web profile page)
galagos auth login --base-url https://app.galagos.ai/api

# 2. Pick a sandbox and a chat
galagos projects ls
galagos projects use <project-id>
galagos chats new --title "BLAST run from CLI"

# 3. Talk to the agent
galagos chat "Run BLAST on workspace/queries.fasta against nr"

# 4. Read results
galagos files cat results/hits.tsv
galagos files get results/hits.tsv -o ./hits.tsv

Commands

Command	Purpose
galagos auth login | logout | whoami	Manage API token and base URL
galagos projects ls | use <id>	List projects, set the default
galagos chats ls | new | use <id>	Manage threads inside a project
galagos chat "<msg>" [--resume] [--tools]	Send a message and stream the reply
galagos chat cancel	Cancel the current turn
galagos files ls | cat <p> | get <p> -o F | put <local>	Browse and transfer files

--project / --thread override the saved defaults on any command.

Configuration

Stored at ~/Library/Application Support/galagos/config.toml on macOS (or the equivalent XDG path on Linux). The token is written with 0600 permissions.

Environment overrides:

• GALAGOS_API_URL — base URL (e.g. http://localhost:8000/api)
• GALAGOS_TOKEN — bypass the saved token

Streaming and reconnection

galagos chat consumes the same SSE stream the web UI does. If the connection drops, the agent keeps running server-side; reconnect with:

galagos chat --resume

Auth note

Phase 1 reuses the existing per-user launcher token (DRF authtoken.Token). One token per account, no expiration. A named/scoped API-key system is planned for Phase 2.