advai-cli
One advai command for local skills, Chrome extension browser control, external CLIs and a terminal chat with a model
- 278 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- advai browser exec runs JavaScript on pages in your signed-in Chrome
- Browser automation reads network events and takes screenshots of open sites
- The program installs third-party CLIs and skills and syncs them into agent directories
Install
- Works in
Manual install
pip install advai-cliRequires Python 3.10 or newer. After install run advai --help and advai info.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- advai browser exec runs JavaScript on pages in your signed-in Chrome
- Browser automation reads network events and takes screenshots of open sites
- The program installs third-party CLIs and skills and syncs them into agent directories
This is third-party code. Review the repository files before installing.
What it does
advai-cli is a Python command-line program with a single `advai` entry point that combines five jobs. The first is local skills: install, list, info, update and uninstall, plus syncing installed skills into agent directories such as Cursor, Claude Code, Codex and TRAE. Skills come from GitHub, the skills.sh registry and other sources, and the state lives in ~/.advai/skills. The second job is the browser. The `advai browser` command opens pages in named sessions, runs JavaScript, waits for a selector, clicks, types, reads network events and takes screenshots. A Chrome extension from the Chrome Web Store provides the live connection, and a local daemon starts on its own. Work happens in a browser where you are already signed in to the sites you need. The third part is external CLIs. The program finds and installs supported console tools through the OpenCLI ecosystem and passes through commands like `advai cli <name> ...`. The fourth is the terminal chat `advai tui` with a configurable model, API base URL, system prompt and transcript export. The fifth is local knowledge bases: add documents, search, resync from source files.
Who it is for. For developers who want to manage skills, the browser and external CLIs from one command and keep state local.
Good fit when
- You need to spread skills across several agents with one command instead of copying folders
- An agent has to open a page in your already signed-in Chrome and pull data out with `advai browser extract`
- You want to try a model from the terminal over any OpenAI-compatible API
- You need a local knowledge base: add documents, search them and resync when files change
- You use third-party CLIs from the OpenCLI ecosystem and want a single entry point for them
Not a fit when
- You need automation without the Chrome extension: browser control relies on the extension and a local daemon
- You do not want to give an agent access to your signed-in browser: `exec` commands run JavaScript on open pages
- You only need a skill manager: the program does more and installs extra components
- You cannot run Python 3.10 or newer: the core is written in Python
Example request
Open https://example.com through advai browser in a session called demo, wait for the h1 heading and extract the textLimitations
- Python 3.10 or newer is required. The npm package creates its own Python virtual environment during postinstall and installs advai-cli from PyPI there, so Python is needed even with an npm install.
- `advai browser` needs the Chrome extension and a first-run setup. Without it `advai browser doctor` reports that support is missing.
- Discovering and running external CLIs needs the opencli program, and the chat needs an ADVAI_API_KEY key and an OpenAI-compatible server.
- The current release uses a fixed internal extension context for browser routing.
How to disable. Uninstall the package: pip uninstall advai-cli, npm uninstall -g advai-cli or brew uninstall advai-cli. Delete the ~/.advai directory with skills, knowledge bases and browser state by hand and disable the Chrome extension in the browser.
FAQ
How do I install advai-cli?
Run pip install advai-cli from PyPI. There is also npm install -g advai-cli, brew tap Advai-X/tap followed by brew install advai-cli, and the install.sh script. Check the install with advai --help and advai info.
How do I connect a browser to advai?
Install the advai-cli extension from the Chrome Web Store, open it once for first-run setup and sign in to the sites you need. Then run advai browser doctor: it shows the daemon state and connected extensions.
Which agents does advai sync skills into?
Into platform directories of agents such as Cursor, Claude Code, Codex and TRAE. The advai skill list command shows installed skills and advai skill info <name> shows metadata.
Which model does the terminal chat use?
Any model behind an OpenAI-compatible API. advai tui lets you configure the model, API base URL, system prompt and transcript export, and the key comes from the ADVAI_API_KEY variable.
Where is the state stored?
Locally in ~/.advai: skills live in ~/.advai/skills and browser state in ~/.advai/browser.
Discussion
No questions or reviews yet.
Related
CLI toolsA plugin and CLI catalog for agents: generates command-line interfaces for GUI apps like GIMP and Blender and installs ready ones via CLI-Hub
Hycli turns a website URL and a goal into a typed CLI and MCP tools that run in your own browser session, with no cookie leaks.
A Rust CLI for macOS: an agent reads any app's interface through the accessibility tree and clicks, types and scrolls by element refs, with no screenshots
Bitrix24 as mission control for AI dev agents: a kanban task conveyor, shared memory and one-button phone acceptance
Turns one run of a task into a program an agent calls as a single tool, checking the record before and after the save
Ui.Vision RPA
Ui.Vision
An MCP bridge and browser extension for RPA: an agent writes and runs macros that click by DOM, by OCR text and by image in the browser and on the desktop
Comparison with similar tools
Cards that share the main search query, and how they differ.
README in short
The page starts with reasons to choose the tool and an architecture with three distribution paths. Requirements, install, quick start and a browser section with a command list follow. The remaining sections cover common commands, the terminal UI and a development guide.