CLI
oo-cli is OOMOL's command-line interface and the fastest way to consume published OOMOL packages and Cloud Function workloads from terminal-based environments.
It works well in:
- Codex
- Claude Code
- Local terminal sessions
- Individual user workflows that do not need a custom app integration
If you want to search for a package, inspect what it does, run a cloud function, and get the result back quickly, start with oo-cli before considering API or MCP integration.
Why Start With CLI
- Fastest path to value: install once, log in, and start running packages
- Better for AI coding tools: works naturally inside Codex and Claude Code
- Good for end users too: no need to build an app or wire a backend first
- Lower setup cost than MCP: no server wrapper or protocol setup required
- Higher-level than raw API: common flows such as search, inspection, task run, and waiting for results are already packaged as commands
Typical Workflow
- Install the CLI.
- Log in to your OOMOL account.
- Search packages with natural language.
- Inspect the package you want to use.
- Run a block as a cloud task.
- Wait for the result or fetch it later.
Install
The current oo-cli README documents installation with Bun:
bun install -g @oomol-lab/oo-cli
Then sign in:
oo login
Search, Inspect, Run
Search for a package:
oo search "generate a QR code"
Inspect a package:
oo package info foo/bar@latest
Run a cloud function block:
oo cloud-task run foo/[email protected] --block-id main --data '{"text":"OOMOL"}'
Wait for the task to finish:
oo cloud-task wait <task-id>
Read the result later:
oo cloud-task result <task-id>
Codex And Claude Code
oo-cli is designed to work well with AI coding environments. The project README shows the primary usage pattern as:
- install
oo-cli - run
oo login - use
$oo ...from Codex
The CLI also bundles a Codex skill. On first launch, or after oo skills install, it installs the managed oo skill into the local Codex skills directory when Codex is available.
When To Use API Or MCP Instead
Use API & SDK when you are embedding Cloud Function calls into your own application, backend service, or automation code.
Use MCP when you need to expose OOMOL capabilities through a standard MCP server for external agent frameworks or tool ecosystems.