opencli-rs-skill: Give Your AI Agent 55+ Platforms With No API Keys
Prefer to watch? Watch on YouTube · Subscribe for new builds · Read the transcript
So here's the thing that drives builders crazy about AI agents. You set everything up. Your agent is smart, it can code, it can write, it can plan. And then you ask it to check your Twitter mentions or pull a Reddit thread and it just... can't. Because it doesn't have API access.
API keys are a whole thing. OAuth flows, rate limits, paid tiers, terms of service headaches. For some platforms you can't even get API access without a business account. It's frustrating.
A developer named nashsu just dropped a skill called opencli-rs-skill that solves this completely. It gives your AI agent access to 55+ platforms , Twitter/X, YouTube, Reddit, Notion, HackerNews, GitHub, Discord, and way more , by reusing your existing Chrome login session. No API keys. No OAuth. 150+ developers starred it in four days.
I covered this in my latest video and the response was wild. So let me break down exactly what this is, how it works, and whether it's worth installing.
The API Key Problem Nobody Talks About
When people get into building with AI agents, they hit this wall fast. Your agent can browse the open web fine. But most of the interesting stuff , your social platforms, your tools, your dashboards , sits behind authentication.
Getting API access to Twitter costs $100/month for basic access. Reddit's API blew up in 2023. Notion requires building an integration. Most platforms have rate limits even when you do get access. And half the time the data you want isn't even in the API.
The workaround people use? Headless browsers like Playwright or Puppeteer. Those work but they're heavy. You're spinning up a full Chromium instance, dealing with anti-bot detection, managing sessions manually. It's a lot of overhead for what should be a simple task.
opencli-rs takes a different approach. Instead of simulating a browser, it just taps into the one you already have open.
How opencli-rs-skill Actually Works
The core idea is simple. You're already logged into Chrome. You've already authenticated with Twitter, YouTube, Reddit, whatever. Why make your agent start from scratch when it could just reuse that session?
opencli-rs is a Rust-based CLI tool. Single binary, 4.7MB, zero runtime dependencies. It comes with a Chrome extension that bridges the gap between the CLI and your running Chrome instance. When your agent runs a command, it goes through the CLI, through the extension, and executes as if you were doing it yourself in the browser.
The skill wrapper makes this available to Claude Code and OpenClaw as a native skill. So instead of writing custom scripts or managing sessions yourself, your agent can just ask for what it needs in plain language and the skill handles the rest.
Here's the flow:
- You install the opencli-rs binary
- You install the Chrome extension (links your Chrome to the CLI)
- You install the skill in Claude Code/OpenClaw
- Your agent can now access any platform you're logged into
That's it. No API keys. No token management. No rate limit negotiations. If you can access it in Chrome, your agent can access it.
The Full Platform List (55+ and Growing)
This is what got 150 people to star it in four days. The platform coverage is genuinely impressive.
| Category | Platforms | Login Required |
|---|---|---|
| Social | Twitter/X, Reddit, Weibo, Bilibili, Zhihu | Yes |
| Dev Tools | GitHub, StackOverflow, HackerNews, Dev.to, Lobsters, Arxiv | Public + Login |
| Productivity | Notion, Discord, Cursor, ChatGPT | Yes |
| Finance | Yahoo Finance, Xueqiu | Public |
| Media | YouTube, Wikipedia, TikTok, Instagram, Facebook | Public + Login |
| DevOps | Docker, kubectl | Local |
Some platforms work in public mode (no login needed) and some require your session. The README is clear about which is which.
Five Killer Use Cases
So what do you actually do with this? Here are the use cases I'm most excited about.
1. Social Media Monitoring on Autopilot
Tell your agent: "Every morning, pull my top Twitter mentions, check what's trending in my niche on Reddit, and summarize anything I need to respond to." Without this skill, that requires three separate API setups. With it, one instruction and you're done.
2. Research Aggregation
Instead of jumping between HackerNews, Arxiv, Reddit, and Dev.to manually, your agent can pull all of it in one shot. "Find the top discussions about AI agents from the last 48 hours across HN, Reddit, and Dev.to." That used to take 20 minutes. Now it's one message.
3. Content Repurposing
Your agent can pull your own YouTube video stats, check which videos are getting comments, read those comments, and use them to write your next script or post. Closing the feedback loop automatically.
4. Notion as Your Agent's Brain
You probably already use Notion. This skill lets your agent read and write to Notion pages without setting up the API. Your agent can pull your project notes, update task status, and add new entries , all through your existing login.
5. Competitive Intelligence
Tell your agent to monitor a competitor's Twitter, check their GitHub activity, and flag any product announcements. Automate the competitor tracking you keep meaning to do manually.
How to Install It (Step by Step)
This is the part where I want to save you 30 minutes of confusion. Here's the exact order.
Step 1: Install the opencli-rs binary
Go to github.com/nashsu/opencli-rs and follow the install instructions for your OS. Mac, Windows, Linux all supported.
Step 2: Install the Chrome extension
Download it from the GitHub releases page. This is the bridge between the CLI and your Chrome session. Without this, the tool can't read your authenticated sessions.
Step 3: Make sure Chrome is open and you're logged in
The tool reuses active sessions, so you need to be logged into the platforms you want to access. Just have Chrome open in the background.
Step 4: Install the skill
npx skills add https://github.com/nashsu/opencli-rs-skill
Or just paste the GitHub link into your Claude Code or OpenClaw chat and tell it to install the skill. The agent will handle it.
Step 5: Restart Claude Code or OpenClaw
Skills activate on restart. Once you do that, your agent has access to every platform you're logged into in Chrome.
Limitations You Should Know
I'm not going to hype this without being real about the tradeoffs.
Chrome has to be open. This isn't a background service. Chrome needs to be running for the extension to bridge sessions. Not a big deal for desktop setups, but it means you can't run this headlessly on a server.
It acts as you. When it posts to Twitter, it's posting from your account. When it reads your DMs, it's reading your actual DMs. Powerful, but be thoughtful about what you let your agent do autonomously vs. what you want to review first.
Platform terms of service. Using automation on some platforms (looking at you, Twitter) can technically violate ToS. Same as using any browser automation. Judge your risk accordingly.
Session expiry. If you get logged out of Chrome, the skill loses access until you log back in. Not a big deal but worth knowing.
None of these are dealbreakers. They're just things to factor in.
opencli-rs vs. Other Browser Automation Options
| Tool | Setup Time | API Keys Needed | Server-Friendly | Binary Size |
|---|---|---|---|---|
| opencli-rs-skill | 5 min | None | No (needs Chrome) | 4.7MB |
| Playwright | 30+ min | None | Yes | 300MB+ |
| Direct APIs | Hours | Yes (per platform) | Yes | N/A |
| Firecrawl | 15 min | Yes (paid) | Yes | N/A |
For desktop setups where you want your agent to work with platforms you're already logged into? opencli-rs is the fastest path by a mile.
Is This Worth Installing?
If you run OpenClaw or Claude Code on a desktop machine and you're already logged into platforms in Chrome, yes. Install it. The setup takes five minutes and the upside is real.
If you're running agents on a headless server or VPS, look elsewhere. Playwright or direct APIs will serve you better there.
For the rest of us building locally on our Mac or PC, this is one of those skills where you install it and immediately think of ten things you want to try. Which is pretty cool.
The skill is open source. The binary is 4.7MB. No ongoing cost. This is exactly the kind of thing that makes the community around OpenClaw worth being part of.
FAQ
What is opencli-rs-skill?
opencli-rs-skill is an OpenClaw/Claude Code skill that wraps opencli-rs, a Rust-based CLI tool. It gives your AI agent access to 55+ platforms like Twitter/X, YouTube, Reddit, and Notion by reusing your existing Chrome login sessions. No API keys needed.
Is opencli-rs-skill safe to use?
It reuses your existing Chrome session cookies so it accesses platforms as your logged-in account. The binary is 4.7MB with zero runtime dependencies. Review any open-source tool before installing, and be mindful that it can post to platforms on your behalf.
How do I install opencli-rs-skill?
Install the opencli-rs CLI binary from github.com/nashsu/opencli-rs, then install the Chrome extension from the same releases page. Then run npx skills add https://github.com/nashsu/opencli-rs-skill and restart Claude Code or OpenClaw.
What platforms does opencli-rs-skill support?
55+ platforms including Twitter/X, YouTube, Reddit, HackerNews, StackOverflow, Notion, Discord, GitHub, Wikipedia, Arxiv, Dev.to, Bilibili, Zhihu, Weibo, Yahoo Finance, ChatGPT, Cursor, Docker, kubectl, and more.
Does opencli-rs-skill work with Claude Code?
Yes. It was built specifically for Claude Code, OpenClaw, and similar AI agent frameworks. Install via npx skills add and it integrates as a native skill your agent can call naturally in conversation.
What's the difference between opencli-rs-skill and Playwright?
Playwright opens a full headless browser and handles anything. opencli-rs is a lightweight 4.7MB Rust binary that reuses your existing Chrome session. It's faster and simpler but requires Chrome to be running. Playwright is better for server deployments. opencli-rs is better for desktop setups where you're already logged in.
Start Building With Your Agent Today
This is exactly what I talk about in Shipping Skool. The skills, the tools, the real workflows that make AI agents actually useful. Not theory. Not demos. Real stuff you can set up today.
We do six live bootcamps every week. Real builders, real problems, real solutions. If you want to go further with AI agents , whether you can code or not , come join us.
We'll see you in the next one. Be blessed.
Chapters
Transcript
Read the full transcript
0:00This skill for your AI agent gives it access to 55 different websites without the use of an API key. So, what we're talking about is no authentication, no developer accounts, no headaches, and you get access to these websites without the use of an API key all through your Chrome browser. So, you just install this one skill. I have it linked down below. It is on GitHub right now. And your Open Claw agent can browse Twitter, search YouTube, scroll Reddit, check stock prices on Yahoo Finance, read Hacker News, post on social media, and we are just scratching the surface here. 55 plus platforms all from natural language commands. And this thing has already gotten over 150 stars in just 4 days. So, people are already excited
0:47about. This fills a huge demand and need from having to go to all these websites, get the API key, and give them to your AI agent. This solves that problem. I'm going to explain why this matters so much. So, if you've ever tried to connect your AI agent to like any social media platform or third-party website, it's a pain in the butt. You have to go sign up for a developer account, then you have to apply for API access, which if you're using like the TikTok API, that could take months. So, it could take a while, which we don't have time to wait in this AI space. Then after that, you have to figure out the OAuth tokens and refresh tokens, and then the rate limits of all this technical stuff that honestly just makes you want to pull your hair out and be done with it. Don't even get me started on the Twitter
1:32API. That's a huge other video we could make, you know, a whole in video about that. It's just the fact that this fills a demand is a beautiful thing. I highly suggest taking out the repo down below. You know, and by the time you're done getting through all of this with connecting all these different APIs, most of these APIs have strict rate limits. So, if you do get approved and you do get to use the app, you get limited. So, you can make only certain a number of requests per hour or per day, and some platforms don't even have a public API at all. So, getting programmatic access to Instagram or TikTok can take a while. So, this developer, Nashu, came up with a completely different approach. Instead of using APIs, this skill uses your
2:17Chrome browser's login session. So, if you think about it, when you're logged into Twitter in a Chrome browser, your browser already has session cookie that proves you're you. This skill basically says, "Hey, I'm going to reuse that session cookie to make requests as you." No API key needed because you're already authenticated through the browser. It's elegant, and it's built in Rust, so it's blazingly fast. The entire binary is 4.2 megabytes with zero runtime dependencies. You just download it, and it works. I'm going to walk you through what platforms we're talking about here, because the list is 55 different websites, and it would take a while to go through them all, but it covers all the main ones. For social media media, you have X. You have 24 different commands through there. You can search,
3:03post, tweet, read your timeline, check trending topics, view bookmarks, like posts, follow people, all through your agent. And that, you know, I'm just saying, be careful with that cuz you don't know when your agent's posting, but that's pretty cool. Real quick, before we keep going, if you're watching this and you want to actually build with some of these tools, not just watch videos about them, then you're going to want to check out our community down below, Shipping School. We have a full Cloud Code course, a full Open Cloud course, and [music] four live bootcamps every single week, where we actually help you get set up from scratch. Like, actually set this thing up, not just watch a tutorial and figure it out by yourself. And we also provide one-on-one coaching, so you could book a call with me, we could share screens, and I can help you get Cloud Code or Open Cloud
3:49running on your machine. That's it, no fluff. I built this community because watching YouTube's only gets you so far. We launched it just 3 days ago and we have over 55 members. You need people around you who are actually building, people who hold you accountable, and coaches who can help you when you get stuck. I'll put the link in the description down below. Get in now before the price goes up. Also, you have Reddit with 15 commands, you know, hot posts, front page, search any Reddit, search any search any subreddit, you can upvote, save comment. He also have Facebook with 10 commands, Instagram with 14 commands, TikTok with 15 commands, and that's just covering social media basics. Then you have content platforms like YouTube. You could search videos and get transcripts.
4:34Hacker News, you could pull top stories, new stories, show stories, ask stories, job listings, dev.to, Lobsters, Stack Overflow, Wikipedia, arXiv, and academic papers. Then you've got news platforms, BBC News headlines, Bloomberg with 10 different commands for markets, economics, tech, politics, Google News, Google Search, Google Trends. And then here's where it gets more wild. You've also got financial platforms like Yahoo Finance for stock quotes. You got Xueqiu, which is a Chinese stock platform. And you've got desktop app control, Cursor, Notion, ChatGPT Desktop, Discord, and even Codex. Your AI agent can control these desktop apps through natural language commands. Ask
5:19Cursor to refactor a function, search your Notion for meeting notes, send a message in Discord, all from OpenClaw. And wait, there's more. There's a bunch of Chinese platforms, too. Bilibili, Zhihu, Xiaohongshu, Weibo, Douban, and We Read. If you're doing any kind of international research or content creation, this opens up a massive amount of data that's was basically impossible to access programmatically before. So, now let me explain how you actually use this because it's surprisingly simple. First, you need to install open cli-rs. It's a CLI tool, command line interface. That's the base binary. Then, you install this skill into your open claw or cloud code setup. There are two ways to do it. You could just tell your AI
6:04agent, "Help me install this skill." and paste in the GitHub URL, and the agent will install it for you. Not very complicated. Or, you could do it manually with npx skills add and the repo URL. Once it's installed, you restart the skill and it's active. So, the one requirement is that Chrome needs to be open and you need to be logged into whatever platform you want the agent to access. So, be careful here. I wouldn't be logged into any banking institutions, just my two cents there. If you want to browse Twitter, you need to be logged in to Twitter or X, whatever you want to call it. That makes sense, right? The skill is basically just reusing your session cookies. It's not hacking in. It's literally just piggybacking on the fact you're already logged in. And then, you just talk to your agent naturally. Say, "Search
6:50YouTube for open claw tutorials." and it does it. You say, "What's trending on Twitter right now?" and it pulls the trending topics. You say, "Get the top 20 stories on Hacker News." and boom, there they are in a nice clean format. This is very simple to use and I like that it's not complicated. You should be using this. So, for example, for finance, you could say, "Check the price of Apple stock." and it pulls it from Yahoo Finance. The agent figures out what open cli command to run and then executes it. It gives you the results. So, let me talk about why this is a huge thing for specific use cases because that's what matters, right? That's the main thing. How do you actually use something like this? So, use case number one, content research, obviously. So, if you're a content creator, your agent can ask and all of these content platforms
7:37and dozens of others in addition to YouTube, Reddit, and Twitter, and then they could find trending topics, popular posts, engagement patterns, all without you spending hours scrolling through feeds. Who does that anymore, anyways? Imagine waking up and your agent has already compiled a list of top trending topics in your niche across five different platforms. It's pretty nice. That's your content calendar done before your first cup of coffee. Use case number two is competitive analysis. Your agent can monitor competitors on Twitter, check their Reddit activity, see what they're posting on LinkedIn, see what they ordered at McDonald's yesterday, track their YouTube content, all automatically. Before this skill, each one of those platforms would need its own API integration. Now, it's just one skill for you. So, use case number three is social media management. Your agent can now post to Twitter, engage
8:23with replies, share content across platforms. If you're managing multiple social accounts for a business, this is huge. Instead of logging into five different platforms every day, your agent handles the distribution. Use case number four is market research. Pulling stock prices from Yahoo Finance, read Bloomberg headlines, scan Reddit investing communities, check Twitter for market sentiment, all in one conversation with your agent. I mean, that is crazy. For anyone in finance or trading, this is like having a research assistant that never sleeps. Use case number five is international research. Those Chinese platforms I mentioned, Bilibili alone has over 300 million active users. Zhihu is basically China's Quora. Xiaohongshu is a basically their
9:10Instagram. If you're building anything that touches the Chinese market or doing research into trends, that is a start in Asia. This skill opens up a world of data that almost was impossible to access before. If you're struggling to keep up with content, well, I'm about to save you about 40 days worth of work. I built [music] something called Content Machine. It's 10 AI agents that run on the Open Claw orchestration, and they handle everything. Scripts, thumbnails, X posts, [music] blogs, outreach, clips, newsletters, all of it. So, I went from 1,000 subscribers to 4,000 subscribers on YouTube in 7 days using this exact system. Every single morning, I wake up and the content's already done. I spend maybe 15, 20 minutes reviewing and
9:56approving them and I move on with my day. It works for [music] any niche, fitness, finance, real estate, marketing, whatever you're building and it is 100% completely customizable to your use case. So, you get [music] the mission control dashboard, all of the cron jobs, everything I've built over the last 40 days helping me gain more and more people to subscribe and join the community. So, you plug in your own thing and it molds it to you, it learns how you talk and it writes so it doesn't sound like AI slop. $97 one time, it's not a subscription. I'll put the link down below and you'll thank me later. Now, I want to re-real about some of this because you think it can't be this easy. There are some limitations. I will always keep it honest and up front with you. First, this does require Chrome to
10:42be open, which means it works best on a desktop machine or a server that has Chrome running. If you're running OpenClaw on a headless VPS without a browser, you need to set up something like a virtual display. That's doable, but that's an extra setup. The second thing is since you're reusing your browser session, anything the agent does on the platforms is happening as you. So, a little scary. So, if you tell it the agent to post a tweet and that tweet is going from your account, if you tell it to post on Instagram, those likes are from your account. So, be thoughtful about what you let the agent do autonomously versus what you review first. I always recommend having an approval step for anything that posts publicly, but whatever you want to do. The third thing is rate limits still technically apply at the platform level.
11:27Twitter might notice if you're making hundreds of requests per a through the browser session. The tool is designed to be reasonable about this, but don't go crazy trying to scrape a million posts. Use it like a human would browse, just faster and more organized. But even with those caveats, this is one of the most practical skills I've seen drop for open claw because the number one complaint I hear from people building agents is, "How do I connect my agent to my X account?" And the answer used to be go set up a API, deal with the tokens, handle the rate limits, write the integration code. Now the answer is just install this one skill and talk to your agent. Honestly, I think it's just the beginning and a great step in the right direction. The repo is open source, which means the community is going to add support for even more platforms over time. Someone's probably already working on adding Pinterest, Threads, WhatsApp,
12:14and who knows what else. The architecture is extensible. Every new platform that gets added makes the skill more valuable for anyone using it. And I mean, come on, 152 stars in 4 days? It's getting traction. That tells you something, right? People are excited about it because it solves a real problem that everybody has been dealing with. And shout out to Nashu. Nashu, shout out. You're doing great, buddy. For anyone building this in Rust instead of Python or JavaScript, I mean, hands off to you. That 4.7 megabyte binary with zero dependencies is exactly how tools should be built. No pip install, no npm install, no dependency conflicts. Just download it and run. So if you want to learn how to set up your own agent with skills like this, that's exactly what we do in the community. We've had 170 people join in 17 days, and it is a huge blessing to work with people in the
12:59community that are geared towards working towards the same thing. So we do six live calls a week. I put a link in the description if you want to be building along other people building AI agents. You know, if you think about where we're even going and where we were maybe 6 months ago, if you wanted your agent to check Twitter, you needed the Twitter API. That cost $200 a month for access. If you wanted Reddit data, you needed the Reddit API. YouTube, you needed the YouTube API. Each one was a separate project, separate credentials, separate code. Now it's just one install in a single conversation. That's what progress looks like. The bottom line is if you're running any kind of AI agent and you want to actually interact with the internet the way you do, the way browse, you know, the way you browse Twitter or check Reddit or search YouTube, this skill is probably one of
13:45the most important skills you need to install this month. One skill that covers 55 platforms, launched 4 days ago, 152 stars, zero API keys. That's it. So if you're looking to build something out like this, you could definitely join the community. I would get to teach you how to integrate something like this. And if you haven't subscribed to the channel already, we release four to six videos every day, dropping GitHub repos, dropping open call use cases, and breaking AI news. We'll see you guys in the next video. Check out that repo down below. Bye-bye.
Ready to start building with AI?
Join the Shipping Skool community free. Paid members get live bootcamp calls every week and weekly-updated courses.
Join Shipping SkoolKeep reading
A Malicious OpenClaw Skill Hid a Base64 Payload, Here Is How to Scan Your Mac in 10 Seconds
OpenClaw CRM: How DenchClaw Runs Your Sales Pipeline Locally
DenchClaw is a local CRM framework built on OpenClaw. You manage your pipeline by talking to the agent, which can import leads, change views, and run browser-assisted sales tasks. Your data stays in a DuckDB file on your own machine. Start with narrow permissions and clear stage definitions.
How to Automate a Cleaning Business with an OpenClaw AI Agent
Tony, a cleaning business owner who doesn't code, used one OpenClaw agent to answer inquiries by text, quote prices from his own pricing rules, book jobs in Google Calendar, send reminders, and ask for reviews. His phone time dropped from 4-5 hours a day to about 45 minutes.
