uptonica

The Uptonica command-line client for developers, agencies, and anyone automating their workspace — one binary, scope decided by the token you configure, not by which command you type.

Install

pip install uptonica   # or: uv tool install uptonica

Log in

uptonica login

Opens your browser, you confirm the code and pick which workspace(s) and areas the token should reach, and the token comes back to the CLI on its own — no copy-paste. (On a machine with no browser, uptonica login --no-browser prints the URL and code to visit instead.)

Prefer to mint the token yourself? Go to Settings -> API Tokens in your Uptonica workspace (https://app.uptonica.com/account/api-tokens), pick "Terminal access", choose scope, copy it, then:

uptonica config set-token

Use it

uptonica whoami
uptonica tenant use my-store          # set a default so you don't repeat --tenant
uptonica tools --search contact       # find a tool without scrolling the whole catalog
uptonica tools --show catalog.product.get   # see its full description + parameters
uptonica call catalog.stats.summary
uptonica call crm.contact.create_task --arg title="Follow up" --arg priority=2
uptonica ask "quanto ho venduto questo mese?"

If your token reaches more than one workspace, --tenant on any command overrides the default for that one call.

A write tool that needs confirmation returns a confirmation_token; re-run the same command with --confirm <token> within ~15 minutes to actually apply it. --dry-run previews a write without executing it. A tool name that doesn't exist gets a "did you mean" suggestion instead of a bare error.

Ask in natural language

uptonica ask "<message>" skips the tool catalog entirely — Lia (the same assistant behind chat and the mobile app) reads your request and picks the tool(s) herself, with the same per-tool permissions she already has everywhere else. A workspace's conversation continues across calls by default, so a follow-up like uptonica ask "sì" answers whatever Lia just asked (she always asks before a write, same as chat). Use --new to start a fresh thread, or --conversation <uuid> to pick a specific one.

Run uptonica with no arguments (or uptonica ask with no message) to open an interactive session instead — one process for the whole conversation, with /workspace to list or switch workspaces, /new for a fresh thread, and /help for the rest. Ctrl+D exits. Enter sends a message; Option+Enter (Alt+Enter) starts a new line without sending. Every tool Lia calls shows up live (◐ catalog.stats.summary... → ✓), so a slow answer never looks like it's just hung.

For AI agents

If you're an agent (Claude Code, Codex, or anything else) about to drive this CLI on someone's behalf, read AGENTS.md first — discovery order, how to handle a multi-workspace token, and the dry-run/confirm flow for writes.

Claude Code users: this repo is also a plugin marketplace, so you don't need to clone anything to get that guidance loaded automatically:

/plugin marketplace add tonic-tech/uptonica-cli
/plugin install uptonica@uptonica-cli

Status

Actively developed — v0.2.2, built directly on the same tool API that powers Uptonica's in-app assistant. See the repo issues for what's still open.