Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

OpenCLI

Turns websites into JSON-returning CLI commands and lets agents drive your logged-in Chrome through a local extension bridge.

Reviewed by Sheela · Updated

View on GitHub Open source

Our rating

3.8/5

OpenCLI is a fast way to give an agent your browser and your logins, which is exactly why it needs treating like a credential rather than a utility.

What is OpenCLI?

OpenCLI, from GitHub developer jackwener, gives AI agents and scripts two ways into the web through your own Chrome. Built-in adapters turn more than 100 sites, including Reddit, Hacker News, X, LinkedIn, Bilibili, Xiaohongshu, Zhihu and Amazon, into deterministic commands such as opencli hackernews top that return JSON. For anything else, opencli browser primitives (open, click, fill, extract, network, screenshot) let an agent operate any page using your existing logins.

It connects through a Browser Bridge extension from the Chrome Web Store and a small local daemon, with an OpenCLIApp tray app for macOS and Windows or an npm package for servers. Agent skills install with npx skills add jackwener/opencli, and one of them guides an agent through writing a new site adapter end to end. Commands use sysexits-style exit codes and a fixed JSON envelope.

The repository was created in March 2026, is Apache-2.0, has about 30,800 GitHub stars, and is at v2.0.2 (20 October 2026); v2.0.0 three days earlier changed CLI syntax and JSON output.

Is OpenCLI worth it?

OpenCLI's useful idea is to make repeat web tasks deterministic. Rather than an agent re-reading a page every time it wants Hacker News's front page or a Bilibili creator's stats, an adapter fetches it through the logged-in browser and returns JSON with a stable envelope and exit codes (66 empty, 69 bridge down, 77 auth required). That is cheaper in tokens and far easier to script than screenshot-driven browsing, and the adapter-authoring skill lets an agent add a site by recon, endpoint discovery and a verify step.

The coverage reflects its origins. The richest adapters are for Chinese platforms (Xiaohongshu has 16 commands, Bilibili 18) alongside X, Reddit and LinkedIn, and several are write commands: post, reply, follow, upvote, publish. Download support pulls media from Xiaohongshu, X, Pixiv and others. Many of those uses sit awkwardly with site terms of service, and that risk falls on the account holder, not the tool.

Security is the main consideration. The agent acts inside your real Chrome profile with every session you are signed into, through an extension and a daemon on localhost:19825. Prompt injection on a page the agent reads can therefore reach accounts it was never asked to touch; profile aliases and per-session tab leases help keep work in a dedicated profile, which is the sensible way to run it.

It is young and moving. v2.0.0 rewrote the command contract days before this review, about 316 issues and pull requests are open, and adapters break whenever sites change, which is why an autofix skill exists. Compared with browser-use, which runs its own browser and suits unattended agents, OpenCLI wins on reuse of existing logins and deterministic commands and loses on isolation.

Where it earns its place

  • A social media manager at a consumer brand selling in China

    Pull weekly creator and note statistics from Xiaohongshu and Bilibili

    Run opencli xiaohongshu creator-stats and opencli bilibili creator-stats on a schedule with -f csv into a spreadsheet. Free; the account's own session does the work, so stay within platform rate limits.

  • A developer using Claude Code for research

    Let the agent read pages behind a login without handing over passwords

    Install the opencli-browser skill and point it at a separate Chrome profile signed into only the needed sites; the agent navigates and extracts through DOM snapshots rather than screenshots.

  • A data engineer maintaining internal scrapers

    Replace brittle one-off scripts with versioned adapters

    Write adapters with opencli adapter init and test, keep them in a plugin repo installed via opencli plugin install, and use opencli-autofix when a site change breaks one. Pin the version, since 2.0 changed the JSON contract.

Before you commit

Setup
Install OpenCLIApp on macOS or Windows, or npm install -g @jackwener/opencli on Node.js 20.18.1+, then add the Browser Bridge extension from the Chrome Web Store and run opencli doctor; first command in a few minutes.
Integrations
Chrome or Chromium via extension, or a remote browser over a CDP endpoint; agent skills for Claude Code, Cursor and other skill-compatible agents; community plugins from GitHub; yt-dlp for video downloads.
Data and privacy
Runs locally; the daemon listens on localhost:19825 and the extension acts with the cookies of the connected Chrome profile. Adapter knowledge is kept in ~/.opencli. No hosted service is involved; what the agent reads goes to your agent's model provider. Apache-2.0.
Cost at scale
No paid edition. The real costs are a machine running Chrome for each automated profile and the maintenance of adapters as sites change.
Limits
About 316 open issues and pull requests; breaking 2.0 contract change; render-dependent commands can stall on background tabs when pooling; site anti-automation controls may block or flag accounts.

What’s good

  • 100+ site adapters returning stable JSON
  • Uses your existing Chrome logins, no API keys
  • Agent skills to write and repair adapters
  • Clear exit codes and output formats for scripting

What’s not

  • Agent acts with every session in the Chrome profile
  • Write commands and downloads raise terms-of-service risk
  • v2.0 changed syntax and output in October 2026
  • Adapters break when sites change

Best for

Developers and power users who want agents to read and act on sites they are already logged into, from a dedicated Chrome profile.

OpenCLI alternatives

OpenCLI: common questions

Is OpenCLI worth it?
OpenCLI is a fast way to give an agent your browser and your logins, which is exactly why it needs treating like a credential rather than a utility. We rate it 3.8 out of 5.
How much does OpenCLI cost?
Free and open source (Apache-2.0), including the Chrome extension and desktop app.
What are the downsides of OpenCLI?
Agent acts with every session in the Chrome profile; write commands and downloads raise terms-of-service risk; v2.0 changed syntax and output in October 2026; adapters break when sites change.
Who is OpenCLI best for?
Developers and power users who want agents to read and act on sites they are already logged into, from a dedicated Chrome profile.
How much does OpenCLI cost for a team or in production?
No paid edition. The real costs are a machine running Chrome for each automated profile and the maintenance of adapters as sites change.
Is OpenCLI safe for company data?
Runs locally; the daemon listens on localhost:19825 and the extension acts with the cookies of the connected Chrome profile. Adapter knowledge is kept in ~/.opencli. No hosted service is involved; what the agent reads goes to your agent's model provider. Apache-2.0.
What are the best OpenCLI alternatives?
The closest alternatives we have reviewed are Browser Use, Agent Reach and Stagehand.

Answers checked Oct 31, 2026.