If you have used an AI chatbot for Linux work, you have probably had it tell you exactly which nmcli command to run, and then you still had to open a terminal and type it yourself.

The assistant understands the task, but it has no way to reach into the system and carry it out, so you end up doing the last and most important step by hand.

That gap is exactly what openKylin 3.0 is trying to close, which is an open-source Linux distribution developed in China, incubated and operated by the OpenAtom Open Source Foundation, and this release puts an AI agent inside the operating system itself.

It introduces KylinBot, an AI agent framework, and Xiao K, an AI assistant that can control system settings like Bluetooth, Wi-Fi, display brightness, and input devices via voice or text commands.

What makes this more than another chatbot is the way the agent talks to the desktop, means agents call desktop capabilities through the MCP protocol instead of simulating clicks, where MCP (Model Context Protocol) is an open standard that lets an AI model call tools in a structured and predictable way.

In this review, we will look at how that works, install the release, test the agent the way a sysadmin would test any new automation tool, and then decide whether it deserves the “smartest AI Linux desktop” label.

How openKylin 3.0’s AI Agent Works

Before installing anything, it’s worth understanding how the AI pieces connect to the system, because that explains both what the agent can do well and where you should be careful with it later in the review.

KylinBot is the agent framework that does the reasoning, and Xiao K is the graphical assistant you type or speak to. According to the project, KylinBot can understand requests through voice and text, keep context across multiple rounds of conversation, and use system capabilities to carry out tasks.

To reach those capabilities, it integrates Kylin-CUA, a computer-use automation tool that gives it access to system settings and device management.

The desktop side of the connection is handled by UKUI (Ultimate Kylin User Interface), openKylin’s own desktop environment. UKUI 4.24 standardizes the APIs across its modules and integrates KACP (Kylin Agent Control Protocol), so KylinBot can reach underlying desktop capabilities through unified APIs.

The layout below shows how a request travels from you down to the system.

Because the agent works through these APIs, openKylin can let other agents plug into the same layer such as KylinBot, WorkBuddy, OpenClaw, and Raccoon Work can all run on openKylin 3.0, and the project says third-party agents get the same system access as its own.

Each specific task the agent can perform is packaged as a skill, which is a small, self-contained description of one job and the system interfaces it needs, and we will look at those skills directly in the hands-on section.

What’s New in openKylin 3.0 Beyond the AI Agent

The AI agent sits on top of a heavily updated base system, and those lower layers matter just as much if you plan to build software or run scripts on this release.

• openKylin 3.0 moves the kernel from Linux 6.6 straight to Linux 7.0, along with GCC 15, glibc 2.42, LLVM 22, and JDK 25, and more than 180 core components upgraded around the new kernel.
• The release integrates the three NIST post-quantum cryptography standards (ML-KEM, ML-DSA, and SLH-DSA) through the openHiTLS cryptography suite.
• With UKUI 4.24, you can make a fist to take a screenshot, wave your hand to turn pages or adjust volume, and use global voice input to issue commands, and Kylin uses the webcam to capture those hand gestures.
• kylin-time replaces the legacy time utility, kylin-wget replaces the classic wget, and kylin-ki18n replaces ki18n, all rewritten in Rust to reduce memory-safety bugs such as buffer overflows.

We will check how these replacements are installed in Example 4, since a rewrite can behave slightly differently from the tool your scripts expect.

Install openKylin 3.0 or Upgrade From the Beta

Now that you know what the release contains, the next step is getting it onto a machine, and the first thing to plan for is the download size. The ISO is a little over 7 GB, mainly because the OS comes with pre-installed AI models, so you will need a USB stick of at least 8 GB and some patience on a slower connection.

For a first look, a virtual machine is the safest place to start, and we gave ours 4 CPU cores, 8 GB of RAM, and 60 GB of disk, since the bundled models and the agent run alongside the desktop.

Keep in mind that a virtual machine has no real Wi-Fi or Bluetooth radio, so the network and Bluetooth skills are best tested on a laptop, which is why Example 2 later runs on real hardware.

If you plan to install on bare metal, check your hardware first, because the project publishes hardware compatibility lists for x86, ARM, LoongArch, and RISC-V in its download page.

Verify and Write the ISO to a USB Drive

Once you have downloaded the x86_64 desktop ISO from the openKylin download center, it’s a good idea to confirm the file isn’t corrupted before writing it, since a damaged 7 GB download usually fails halfway through the installer with no clear error.

Generate its checksum with sha256sum and compare the long string in the output with the SHA256 value published on the download page.

sha256sum openKylin-3.0-*.iso

With the ISO verified, find your USB drive’s device name with lsblk and write the image with dd.

lsblk
sudo dd if=openKylin-3.0-*.iso of=/dev/sdX bs=4M status=progress oflag=sync

Replace /dev/sdX with your USB device, such as /dev/sdb. If you prefer a graphical tool, balenaEtcher or GNOME Disks do the same job.

During installation, we created the user tecmint and placed the VM on a host-only network at 192.168.56.20, which lets us run checks from the host later if needed.

Upgrade an Existing 3.0 Beta Installation

If you are already running the 3.0 Beta, you can skip the reinstall entirely, because openKylin uses the familiar APT and dpkg package tools and the project documents a standard APT upgrade for Beta users:

sudo apt update
sudo apt upgrade

The Update page in System Settings does the same thing if you prefer the graphical route.

Testing openKylin 3.0 From a Sysadmin’s Point of View

With the system installed, we can start testing, and the approach here is the same one you would use for any new automation tool, which is to give it a task and then confirm the result independently from the terminal.

The first example doesn’t involve the AI at all, since it’s worth confirming the base system before trusting anything built on top of it.

Example 1: Confirm the Linux 7.0 Kernel and Toolchain

Open a terminal on the freshly installed system and check the release and kernel version:

cat /etc/os-release
uname -r

The os-release file should identify the system as openKylin 3.0, and the kernel string from uname -r should start with 7.0. The exact suffix after that varies with the build and any updates you have applied, so don’t worry if it doesn’t match another system exactly.

Since the new toolchain matters for anyone compiling software, check the compiler and C library next:

gcc --version
ldd --version

Look for GCC 15 on the first line of the gcc output and glibc 2.42 on the first line of the ldd output. This has one practical consequence worth remembering, which is that a binary you compile here and copy to an older distribution may refuse to start with an error like version 'GLIBC_2.42' not found, because programs linked against a newer glibc can’t run on systems with an older one.

Example 2: Let Xiao K Change Network and Bluetooth Settings

The first example confirmed the foundation, so now we can test the feature that sets openKylin apart. For this one, we moved to a laptop, because the skills involved need real radios. Among the built-in skills, there is deep integration with NetworkManager to manage Wi-Fi, Ethernet, and VPNs, plus Bluetooth toggling, device discovery, and automatic audio device pairing.

Before asking the assistant to do anything, record the current state of both radios, so you have something to compare against once the agent has finished:

nmcli radio wifi
bluetoothctl show | grep Powered

Now open Xiao K and type a plain request such as “Turn off Bluetooth and switch Wi-Fi off“. Once the assistant reports that it has finished, run the same two commands again.

If the agent did its job, nmcli radio wifi now prints disabled and bluetoothctl reports Powered: no. It’s worth doing this check every time you try a new skill, because an assistant saying it changed something and the system actually changing are two separate things, and the terminal is the reliable witness.

If you are testing in the VM instead, you can try the same idea with the wired connection by asking Xiao K to disconnect it and then running nmcli device status. Do this from the VM’s own console rather than over SSH from 192.168.56.1, since disconnecting the network will also cut your SSH session.

Example 3: Explore and Customize Skills From the openKylin-skills Repository

The agent in Example 2 relied on built-in skills, and since those skills are shared in a public Git repository, you can read how they work before letting the agent run them on your system.

Clone the repository from Gitee and look at its layout:

git clone https://gitee.com/openkylin/openkylin-skills.git
cd openkylin-skills
ls

Each skill describes what the agent is allowed to do and which system interfaces it calls, so this is the best place to learn the format before writing your own.

The project encourages this, since developers can build skills with a unified AI SDK and submit them to the openKylin-skills repository. One limitation to expect is that the repository is in Chinese, so you may need a translation tool to follow the skill descriptions for now.

To customize the agent for your own environment, a sensible first skill is something narrow and read-only, such as reporting disk usage with df -h or listing failed services with systemctl --failed.

Starting with read-only skills lets you learn how the agent interprets requests before you give it anything that can change the system, in the same way we verified its changes in Example 2.

Example 4: Check Which Tools Were Replaced by Rust Versions

The earlier examples focused on the AI features, but the Rust rewrites mentioned earlier can quietly affect scripts you bring over from other distributions. To see what wget actually points to on this system, follow any symlinks to the real binary and then ask dpkg which package owns it:

readlink -f "$(which wget)"
dpkg -S "$(readlink -f "$(which wget)")"

If the owning package is kylin-wget rather than the classic wget package, test your download scripts carefully, since a rewrite can differ in rarely used options and in the exact wording of error messages that some scripts parse.

Verdict: Is openKylin 3.0 the Smartest AI Linux Desktop Yet?

After testing the base system, the agent, and the skills, the answer depends on what you mean by “smartest“. For letting an AI agent change real system settings through proper desktop APIs, openKylin 3.0 is the most complete option available on Linux right now, and the fact that third-party agents get the same access makes it more open than a single-vendor assistant.

The weaker points are practical ones. The skill ecosystem is mostly documented in Chinese, the 7 GB ISO is heavy, and the Token Center login raises fair questions about where your prompts go.

For privacy-focused users or production servers, a mainstream distribution with a local model that you control is still the safer choice, while openKylin 3.0 is well worth a weekend on a spare laptop if you want to see where desktop AI is heading.

We’d love to hear how you would use an AI agent that can change your system settings. Would you give it access to network and Bluetooth controls on a work laptop, or keep it limited to read-only skills?

Have you tried writing a custom skill for KylinBot yet? And if you installed openKylin 3.0 on real hardware, feel free to share your machine model and any errors you ran into in the comments, so other readers can learn from them.

The post openKylin 3.0 Review: Is This the Smartest AI Linux Desktop Yet? appeared first on WEBNET Hosting.

The best AI for solopreneurs includes Hostinger Agent for covering several business functions, ChatGPT for general assistance, Perplexity for source-backed research, and Zapier for straightforward workflow automation. Notion AI and Fathom fill more specialized roles in operations, meetings, long-form work, and customizable automation.

Each covers a different part of running a one-person business, so the best choice depends on where you need the most support.

Here are the best AI tools for solopreneurs compared side by side:

1. Hostinger Agent – Best for managing multiple business functions

Hostinger Agent is a business-focused platform with seven specialized assistants: Business Advisor, Creative Writer, SEO Consultant, Marketing Planner, Legal Advisor, Customer Comms, and Sales & Outreach. That breadth suits solopreneurs who need structured support across several business functions without having to open a different general chatbot for each task.

If you’re launching a new service, you can use it to develop the offer and pricing, then carry that work into keyword research, website copy, lead generation, and outreach.

Because the work stays in the same conversation, Hostinger Agent can use the context from earlier stages as you move through the project.

More than 150 guided skills turn common business tasks into step-by-step workflows, including business planning, audience definition, SEO audits, campaign creation, outreach, and customer communication.

Select what you want to accomplish and follow the workflow to provide the relevant information and produce a usable starting point.

Hostinger Agent also connects to more than 1,000 apps, including Gmail, Google Sheets, Notion, HubSpot, and Google Calendar.

With your approval, it can use information from connected apps to add leads to a CRM, update a Google Sheet, create calendar events, or send follow-up emails.

Hostinger Agent strengths

• Keeps context as you move through different stages of a project, making it useful for workflows spanning several business functions.
• Guided skills provide structure for unfamiliar tasks, so you don’t need to know every step of the process before getting started.
• Can carry approved tasks into connected apps, which reduces manual handoffs between researching, creating, organizing, and acting on information.
• Supports recurring tasks and scheduled jobs, taking routine work off your daily to-do list.

Hostinger Agent limitations

• Important outputs still require human review, especially legal documents, pricing recommendations, business strategy, factual claims, and sensitive customer communications.
• The 1,000 monthly credits can limit how much you use the agent, particularly if it becomes part of several recurring business workflows.
• Its broad business coverage may be less suitable when you need advanced features from a specialized tool, such as dedicated CRM, SEO, or analytics software.

Hostinger Agent pricing

New users can try Hostinger Agents with 200 credits. Paid plans start at $6.99/month with a 12-month term. The three-month option costs $7.99/month, while the monthly option costs $9.99/month. Plans include 1,000 credits per month, 150+ skills, and access to 1,000+ app integrations.

Best for: Solopreneurs building anAI tool stack to support planning, marketing, lead generation, sales, and customer communication across a consulting, freelance, coaching, agency, or other online service business.

2. ChatGPT – Best general-purpose AI assistant

ChatGPT is a general-purpose AI assistant that can help you plan, research, analyze information, solve problems, and create business materials through conversation.

Starting a consulting business, for example, could involveusing AI for business planning, then uploading your customer research to identify common needs, compare service ideas, work through pricing options, and turn your decisions into a launch plan.

For ongoing work, Projects let you keep related chats, files, and instructions together. You could create one for your business and add customer research, product or service information, brand guidelines, and other reference materials.

ChatGPT can then use that context when you return with a new task, such as analyzing customer feedback or planning your next campaign.

You can also use ChatGPT for work that requires more than a standard conversation. Deep research can gather and synthesize information from multiple web sources, while data analysis can examine uploaded spreadsheets and turn the data into calculations, tables, or charts.

For tasks you do regularly, you can create a custom GPT that follows the same instructions each time. Scheduled tasks can automatically run prompts at set times, such as preparing a weekly business summary or reminding you to follow up with leads.

ChatGPT strengths

• Works well for tasks that develop through discussion, allowing you to question assumptions, compare options, and refine an idea over several exchanges.
• Projects give ongoing business work a shared context by keeping relevant conversations, files, and instructions together.
• Can work directly with documents, spreadsheets, images, and other uploaded information, which is useful when you need to analyze your own business data.
• Custom GPTs and scheduled tasks can turn instructions you use repeatedly into reusable or recurring workflows.

ChatGPT limitations

• You need to provide enough business context and clear instructions to get specific, useful responses. Vague prompts can produce generic recommendations.
• Important facts, calculations, recommendations, and other generated information should be checked before being used to make business decisions.
• Some capabilities have lower limits on the Free plan, including file uploads, data analysis, deep research, and memory.

ChatGPT pricing

ChatGPT has a free plan with access to its core AI model, web search, file and image uploads, data analysis, and image generation, although some features have usage limits. Paid individual plans start at $8/month for Go, $20/month for Plus, and $100/month for Pro.

Higher tiers offer higher usage limits and expanded access to features such as advanced reasoning, in-depth research, memory, and data analysis.

Best for: Freelancers, consultants, coaches, creators, and other solopreneurs who need a flexible assistant for research, planning, analysis, problem-solving, and everyday business tasks.

3. Perplexity – Best for research

Perplexity is an AI-powered search engine that browses the web, combines information from multiple sources, and returns answers with citations.

Say you’re considering adding a new service to your business. You can use Perplexity to research demand, identify competitors, compare how they position and price similar services, and investigate market trends.

You can then follow the citations to check the evidence behind the findings before deciding whether the opportunity makes sense for your business.

This source-first approach is also useful when you’re entering an unfamiliar market or need to get up to speed quickly on a topic.

You can ask follow-up questions to narrow the research, compare information across sources, and investigate individual claims without manually starting a new search each time.

For more complex questions, Perplexity’s Research mode can carry out multi-step research and produce a more comprehensive report.

This can help with tasks such as comparing several competitors or investigating a market from multiple angles.

Perplexity strengths

• Provides citations with its answers, making it easier to trace information back to the original source and verify important claims.
• Synthesizes information from multiple web sources, which is useful when comparing competitors, markets, products, or industry developments.
• Supports follow-up questions, so you can progressively narrow a broad research question as you learn more about the topic.
• Research mode can investigate more complex questions that require multiple searches and sources.

Perplexity limitations

• A citation doesn’t guarantee that the source is reliable or that Perplexity has interpreted it correctly, so you still need to inspect important sources yourself.
• Research findings can provide evidence for a business decision, but you still need to judge how relevant that information is to your customers, market, and circumstances.
• The free plan is useful for light research but currently limits advanced usage to three Pro Searches per day and one Research query per month. Frequent competitor or market research may therefore justify a paid plan.

Perplexity pricing

Perplexity has a free Standard plan with basic searches and limited access to advanced research features.

Paid individual plans include Pro at $20/month and Max at $200/month. Pro increases access to Pro Search, Research, advanced AI models, and file analysis, while Max provides higher limits for intensive use.

Best for: Solopreneurs launching into a new market, developing a new offer, or regularly researching competitors and industry developments.

4. Notion AI – Best for operations and knowledge management

Notion AI brings AI into the same workspace where you manage projects, notes, documents, databases, and other business information. This makes it useful for solopreneurs who use Notion as a central hub for organizing their work.

Say you manage client projects in Notion. After a client call, Notion AI can summarize your meeting notes and identify action items. You can then use those notes alongside the existing project information to draft next steps, update project documentation, or prepare a follow-up.

It can also help you retrieve information you’ve already saved. If you need to remember what a client requested several weeks ago or find a decision buried in project notes, you can ask a question and have Notion AI search the relevant workspace content for the answer.

This becomes more useful as you accumulate customer research, project documentation, content plans, and other business knowledge.

Beyond individual tasks, Notion includes AI features for recurring operations. AI Meeting Notes can capture and summarize meetings, while database automations can process data when changes occur in a database.

Higher-tier plans also include Custom Agents for recurring workflows, such as compiling project updates or organizing new information.

Notion AI strengths

• Works directly with the notes, documents, databases, and project information already stored in your workspace.
• Makes it easier to find and summarize information across a growing collection of business pages.
• Can turn meeting notes and project updates into summaries, action items, and follow-up work.
• Combines AI with databases and automations, which is useful for keeping recurring operational work organized.

Notion AI limitations

• It is only useful if you already use Notion, or plan to centralize a meaningful part of your business there.
• Notion AI has usage allowances even on the Business plan. Some features can temporarily pause if you reach the rolling six-hour or monthly allowance.
• Custom Agents are not included in the regular AI allowance. They use separate Notion credits based on how much information they process, how many steps they perform, and how often they run, so recurring automations can add to your monthly cost.

Notion AI pricing

Notion has a free plan for individuals that includes basic forms and sites, Notion Calendar, databases with subtasks and custom properties, and a limited trial of Notion AI. Plus costs $9.50/member/month, and Business costs $19.50/member/month when paid monthly. The Business plan includes Notion Agent, AI Meeting Notes, Enterprise Search, and Research mode. Custom Agents use separate Notion credits at $10 per 1,000 credits.

Best for: Solopreneurs who already manage projects, notes, client information, content plans, or business documentation in Notion and want AI to work directly with that existing workspace.

5. Fathom – Best for meetings and follow-ups

Fathom is an AI meeting assistant that records and transcribes calls, creates summaries, identifies action items, and lets you search across past conversations.

For a solopreneur, that can be useful during sales calls, discovery calls, client meetings, or interviews.

Along with otherAI tools for sales, Fathom can reduce the work around these conversations by capturing what was discussed and turning it into summaries and action items you can use for the next step.

For example, after a client meeting, you can review the transcript, pull out commitments and next actions, and use the generated follow-up email as a starting point.

Fathom also lets you search meetings by attendee or keyword and use Ask Fathom to find specific details from any previous call.

Its integrations can extend that workflow beyond the meeting itself. Fathom supports CRM sync and connects with tools such as Zapier, Make, ChatGPT, and Claude, while its public API and MCP support allow meeting information to feed other workflows.

Fathom strengths

• Creates transcripts, summaries, and action items automatically, so you don’t have to take detailed notes during meetings or manually turn them into a list of next steps afterward.
• It allows you to search past calls and find specific decisions, commitments, or discussion points without replaying the full recording.
• Integrates with CRM, project management, automation, and AI tools, which makes it easier to continue the workflow after the call.
• The Free plan includes unlimited recordings and transcriptions, which is useful if you have frequent meetings but do not need advanced features.

Fathom limitations

• You still need to check summaries and action items before relying on them for important commitments, pricing, deadlines, or customer requests.
• Recording meetings can create privacy and compliance concerns, particularly when calls contain customer or confidential business information, and you may need participants’ consent before recording.
• More advanced features, such as AI-generated action items, advanced summaries, and a conversational meeting assistant, require the Premium plan.

Fathom pricing

Fathom has a free individual plan with unlimited recordings, transcriptions, and instant AI summaries.

Premium costs $16/month and adds advanced summaries, AI-generated action items, and the conversational meeting assistant. Team plans start at $15/user/month with a minimum of two users.

Best for: Consultants, coaches, freelancers, recruiters, and other solopreneurs who spend a large part of their week on client, sales, discovery, or interview calls.

6. Zapier – Best for easy workflow automation

Zapier is a workflow automation platform that connects the apps you use so recurring tasks can move from one tool to another automatically.

A Zapier workflow starts with a trigger, such as a customer submitting a form, followed by one or more actions. For example, when a new lead fills out your contact form, Zapier can add their details to your CRM, categorize the inquiry, and create a follow-up task for you.

You can build similar workflows around other repetitive parts of your business. Zapier can turn form responses into tasks, send information between your business apps, categorize customer inquiries, or generate social media copy when new content is published.

AI can be added to individual steps when the workflow needs to summarize, classify, or draft information.

For a solopreneur, the main advantage is that these workflows can connect tools you already rely on. Zapier supports more than 9,000 apps and can also add AI steps for tasks such as summarizing, classifying, drafting, or making simple decisions inside a workflow.

Zapier strengths

• Connects with more than 9,000 apps, so you can automate work between many of the tools you already use.
• Can complete several steps from a single event.
• Can use AI within a workflow to summarize information, categorize requests, or generate text before passing the result to the next step.
• Lets you set conditions for what happens next, such as sending high-priority leads through a different workflow from general inquiries.

Zapier limitations

• Successful action steps generally consume tasks, so complex workflows can use the monthly allowance quickly.
• Automations still need review points when they affect customers, payments, contracts, or other decisions where an incorrect action could have serious consequences.
• Building reliable multi-step workflows takes some setup and testing, especially when several apps and conditions are involved.

Zapier pricing

Zapier has a free plan with 100 tasks per month. The Professional plan starts at $19.99/month and adds multi-step workflows, premium apps, webhooks, and other advanced automation features. Team plans start at $69/month.

Best for: Consultants, freelancers, agencies, and other service-based solopreneurs who use multiple apps and want to automate workflows such as adding new leads to a CRM, creating follow-up tasks, updating customer records, and sending information between tools.

How solopreneurs can use AI

Solopreneurs can use AI across business planning, research, marketing, content, SEO, sales, customer communication, admin, and legal support.

Common applications include:

• Business planning. Use AI to organize ideas, compare options, draft a business plan, test assumptions, and turn a broad goal into concrete next steps. You still decide which direction to pursue, what risks to take, and how much to invest.
• Market and competitor research. Ask AI to gather information about competitors, customer needs, pricing, industry trends, or unfamiliar markets, then compare findings across sources. Check the original sources before relying on the information for an important decision.
• Marketing. UseAI tools for marketing for campaign ideas, promotional copy, audience research, and adapting messages for different channels. You still decide which audiences to target, how to position your offer, and which campaigns are worth pursuing.
• Content marketing.AI for content marketing can help you research topics, build outlines, create first drafts, and repurpose existing content into different formats. Review the final content for accuracy, usefulness, brand voice, and anything that requires your own expertise or experience.
• SEO. Use AI to research keywords, analyze competing pages, identify content gaps, and find opportunities to optimize existing pages. Whenusing AI to improve SEO, review the recommendations against actual search results and your audience before deciding what to create or change.
• Lead generation and follow-up. AI can research prospects, summarize lead information, draft personalized outreach, prepare call notes, and suggest follow-up messages. Review anything customer-facing before sending it, especially when pricing, promises, or sensitive information are involved.
• Customer communication. Use AI to draft routine replies, summarize long email threads, categorize incoming requests, and prepare responses to common questions. Complaints, refunds, unusual requests, and sensitive conversations still need direct human judgment.
• Admin and recurring operations. Automate repetitive tasks such as converting form submissions into tasks, updating records after meetings, preparing recurring reports, or moving information between tools. Start with predictable workflows where you can clearly define the trigger, inputs, and expected result.
• Legal and administrative support. AI can help explain unfamiliar terms, organize requirements, summarize documents, or create a first draft of routine business materials. Treat the result as a starting point and get professional advice when a legal decision could materially affect your business.

Best practices for using AI as a solopreneur

A practical starting point is to automate low-risk backend work first, keep human approval at consequential decision points, protect customer data, and add complexity only when a workflow consistently removes manual work.

Automate backend work where mistakes are easy to spot and correct. Organizing information, summarizing documents, categorizing requests, transferring data between apps, creating first drafts, and preparing recurring reports are good starting points.

Be more cautious with customer communication, financial decisions, contracts, and other actions that can directly affect your business.

Build human approval into the workflow at important decision points. AI can prepare an email, proposal, campaign, analysis, or recommendation, but review it before it reaches a customer, triggers a payment, becomes part of an agreement, or influences a significant business decision.

The same caution applies tolegal AI tools, which can assist with research and documents but shouldn’t be the sole basis for consequential legal decisions.

Be selective about the business and customer data you give AI tools. Before connecting an app or uploading a file, check whether the information is actually necessary for the task. Review the tool’s privacy settings, data retention and training policies, and the permissions granted through integrations. Limit access to the data and actions the workflow genuinely needs.

Give AI reusable context about your business. Create a reference that covers your audience, positioning, preferred terminology, tone, examples of writing you like, and phrases or styles to avoid.

Reusing these instructions across content and communication tasks makes the output more consistent and reduces the need for repeated editing.

Avoid accumulating subscriptions with overlapping capabilities. Before paying for another tool, check whether something in your existing stack already handles the task, how regularly you will use the new capability, whether it can replace another subscription, and whether it connects with your current tools.

Understand a workflow before automating it. Complete the process manually enough times to identify its trigger, required inputs, decisions, expected output, and common exceptions. Then automate the predictable stages first.

This prevents you from spending time building a complex system around a process that is still changing.

Finally, judge an AI workflow by how much work it removes, not simply by whether the automation runs successfully.

Look for reductions in manual steps, context switching, repetitive writing, administrative work, and corrections. If maintaining, monitoring, or fixing the workflow takes about as much effort as completing the task manually, simplify it or stop using it.

For businesses juggling planning, marketing, sales, content, and customer communication, Hostinger Agent can handle several of these functions in a single platform.

If most of your work involves thinking through problems, analyzing information, and creating different types of material, ChatGPT is a broader general-purpose option.

More specific bottlenecks point to more specialized choices. Use Perplexity when research takes up a significant part of your week, Notion AI when the challenge is finding and using information stored across your workspace, and Fathom when meetings create a backlog of notes and follow-up work.

For automation, look at the complexity of the process you want to remove from your workload.

Zapier is a good starting point for straightforward workflows such as moving a new lead from a form into your CRM and creating a follow-up task.

Before adding a second AI subscription, assess how you’re already using AI across your small business and check whether your current tools can meet the new need.

A good AI setup can therefore be as simple as one general tool plus one specialist tool or automation platform.

Add another only when a recurring bottleneck remains, and compare the subscription cost with the time and manual work it actually removes.

All of the tutorial content on this website is subject to

Hostinger’s rigorous editorial standards and values.

The author

Ksenija Drobac Ristovic

Add Hostinger as a preferred source on Google

Ksenija is a digital marketing enthusiast with extensive expertise in content creation and website optimization. Specializing in WordPress, she enjoys writing about the platform’s nuances, from design to functionality, and sharing her insights with others. When she’s not perfecting her trade, you’ll find her on the local basketball court or at home enjoying a crime story. Follow her on LinkedIn.

The post Best AI for solopreneurs: 6 options for running a one-person business appeared first on WEBNET Hosting.

The post I Found these Computers in a Ditch… appeared first on WEBNET Hosting.

AI agents for real estate are AI systems that connect to tools such as your customer relationship management (CRM) platform, email, and calendar to handle tasks for you.

Unlike basic AI assistants or scripted chatbots, they can use live data and choose from approved actions, which helps them handle repetitive tasks that still vary from one client or situation to another.

Some of the most practical uses are lead follow-up, property matching, listing content, client communication, document processing, appointment scheduling, email marketing, and search engine optimization (SEO).

1. Lead qualification and follow-up

AI agents can qualify and follow up with new leads by asking about their budget, timeline, property type, and financing status, then scoring the responses and updating your CRM.

• Needs access to: website forms, CRM, email, calendar, and messaging apps
• Keep human control over: pricing discussions, negotiations, financial or legal questions, and unusual client situations

When inquiries come in through website forms, social ads, open houses, and otherreal estate lead generation methods, an agent can start the qualification process immediately instead of waiting for someone to review the lead manually.

One way to support this workflow is withHostinger Agent, an AI-powered business assistant that connects with Gmail, HubSpot, and Google Calendar. Its sales-focused AI expert can then handle lead follow-up across the tools you already use.

2. Property search and buyer matching

AI agents can match buyers with suitable properties by comparing their requirements with available listings and summarizing why each property fits.

• Needs access to: CRM, authorized listing feeds, internal property data, and buyer requirements
• Keep human control over: listing availability, pricing, property condition, and the final recommendations shared with clients

This can save time when a buyer has several requirements to compare at once. For example, an agent could narrow a large set of listings to properties with three bedrooms, within budget, close to the buyer’s workplace, and with space for a home office.

The quality of those recommendations depends on the data the agent receives. Keeping buyer preferences and other relevant details up to date in aCRM designed for real estate gives it more reliable information to work with.

Matching criteria should stay focused on legitimate property preferences such as price, location, size, and features, not protected characteristics or information that could act as a proxy for them.

3. Listing content creation

AI agents can turn verified property information into drafts for your multiple listing service (MLS), website, email campaigns, and social media.

• Needs access to: verified listing data, brand guidelines, and format requirements for each channel
• Keep human control over: measurements, amenities, location claims, property condition, and fair-housing language

This can reduce the repetitive work of rewriting the same property details for different channels. One verified property brief can become an MLS description, website copy, photo captions, an email highlight, and social media posts, with each version adapted to its format.

A tool such as Hostinger Agent can also use specialized content and marketing AI experts to help prepare different versions of the same verified listing information for multiple channels.

Always check the facts before publishing. AI can produce polished copy from incomplete or incorrect source information, and listing language should describe the property rather than the ideal buyer or neighborhood demographics.

4. Client communication

AI agents can handle routine client communication by preparing replies, reminders, property updates, and follow-up messages.

• Needs access to: email, calendar, CRM, website chat, and messaging apps
• Keep human control over: negotiations, complaints, sensitive conversations, legal questions, and contract discussions

This helps when several clients need updates at the same time. An agent can confirm showings, remind clients about missing documents, notify buyers about new property matches, and prepare post-viewing follow-ups without you drafting each message from scratch.

When communication is spread across several tools, an agent can also use information from connected apps to keep the right context with each client. Hostinger Agent, for example, supports connected business apps alongside a customer communication-focused AI expert.

More sensitive conversations should still stay with you. If a message involves a complaint, negotiation, contract issue, or situation where tone and judgment matter, the agent can prepare a draft for you to review.

5. Document processing and transaction administration

AI agents can help manage transaction documents by extracting key information, organizing files, checking for missing details, and flagging deadlines.

• Needs access to: document storage, transaction management software, CRM, and internal checklists
• Keep human control over: interpreting documents, verifying extracted information, and making legal or transaction decisions

This can reduce the manual work of reviewing and organizing documents throughout a transaction. For example, an agent could scan an inspection report for major findings, compare them with your checklist, and flag an upcoming appraisal deadline.

The agent should support administrative work, not interpret contracts or make legal decisions. Because transaction files can contain sensitive personal and financial information, limit its access to only the documents and data needed for the task.

6. Appointment and showing coordination

AI agents can coordinate property showings by checking availability, suggesting times, sending confirmations, and updating calendars.

• Needs access to: calendar, CRM, email, messaging platforms, and showing management software
• Keep human control over: access requirements, occupied-property rules, restricted viewing times, multi-party arrangements, and sharing access codes

This reduces the back-and-forth involved in scheduling. You can set your available hours and property-specific rules, then let the agent suggest suitable time slots, confirm appointments, update the calendar, and handle routine rescheduling.

For anything involving property access or special viewing conditions, keep the final approval with you.

7. Email marketing and newsletters

AI agents can prepare real estate marketing emails and newsletters by pulling listing information, adding relevant updates, and adapting the content for different audiences.

• Needs access to: CRM, email marketing platform, listing data, website content, and brand guidelines
• Keep human control over: property details, market statistics, personalization fields, recipient lists, and final approval before sending

This can save time when the same campaign needs different versions for different audiences. For example, an agent could create one newsletter for buyer leads, another for seller leads, and a third for past clients using the same underlying listing and market information.

Existingreal estate newsletter formats and examples can provide a starting point, which the agent can then adapt to each audience while keeping the structure and branding consistent.

8. Real estate SEO

AI agents can support recurring SEO tasks by checking existing pages, spotting issues, and preparing updates.

• Needs access to: your website, analytics, search performance data, keyword research, and editorial guidelines
• Keep human control over: search intent, local-market claims, keyword targeting, and final content approval

This can reduce the manual work involved in maintaining and improving a real estate website. For example, an agent could flag pages with missing meta descriptions, suggest relevant internal links, identify search topics for your service areas, or draft updates for older content.

The quality of those recommendations depends on the information you give the agent. Reliable keyword research, accurate website data, and a solidSEO setup for a real estate website give it a stronger basis for identifying useful improvements.

Hostinger Agent also includes an SEO-focused AI expert, making SEO another workflow that can be handled alongside sales, content, and marketing tasks from the same tool.

When to use AI agents for real estate

AI agents make sense for tasks where the next step can be decided from information already available in your tools.

For example, an agent can decide when to send a lead follow-up, which listings meet a buyer’s requirements, or which available time slots to offer for a showing.

They work best when the task:

• Happens often. Lead follow-up, listing updates, showing coordination, and newsletter preparation are all recurring tasks.
• Uses reliable existing data. The agent can work from CRM records, listing feeds, calendar availability, or transaction documents.
• Has clear rules for what happens next. For example, you can tell it to follow up with leads who plan to buy within 90 days.
• Doesn’t depend heavily on human judgment. Drafting, sorting, scheduling, and organizing are better fits than pricing, negotiating, or interpreting legal documents.

Best practices for using AI agents in real estate

To use AI agents safely in real estate, give them accurate data, limit what they can access and do, and keep human review in place for anything that could affect a client or transaction.

• Use reliable, current data. If your CRM and MLS feed show different prices, the agent needs to know which source to trust. Outdated listings, incorrect contact details, or unverified property information can lead to inaccurate results.
• Limit permissions. Give the agent access only to the data and tools it needs for the task. An agent that drafts listing descriptions, for example, doesn’t need access to transaction documents or financial records.
• Keep humans in the loop. Require approval before the agent sends messages, changes important records, or takes actions that could affect a client or transaction.
• Protect client privacy. Transaction files can contain sensitive information such as Social Security numbers, bank statements, and pre-approval letters. Check how connected tools process, store, and share that data before giving an agent access.
• Test before going live. Test realistic situations such as incomplete lead details, conflicting listing information, failed integrations, and messages that should be escalated to you rather than sent automatically.

How to get started with AI agents in real estate

To get started with AI agents in real estate, choose one repetitive, low-risk task and test it before expanding to more complex work. Lead qualification, follow-up messages, listing descriptions, showing coordination, and CRM updates are good starting points. Avoid contracts, negotiations, and other decisions that require significant human judgment.

Connect only the data and tools the task needs, then set clear rules for what the agent can do automatically and what still requires your approval. If you want to start with a ready-made option, Hostinger Agent lets you work with specialized AI experts across several business functions instead of building an agent from scratch.

Test the workflow with realistic situations before using it with clients. Once it consistently handles the task correctly, you can expand it to related work orbuild an AI agent with more complex goals and access to additional tools.

All of the tutorial content on this website is subject to

Hostinger’s rigorous editorial standards and values.

The author

Alma Fernando

Add Hostinger as a preferred source on Google

Alma is an AI Content Editor with 9+ years of experience helping ideas take shape across SEO, marketing, and content. She loves working with words, structure, and strategy to make content both useful and enjoyable to read. Off the clock, she can be found gaming, drawing, or diving into her latest D&D adventure.

The post AI agents for real estate: 8 practical use cases appeared first on WEBNET Hosting.

The post What’s the Worst Tech Linus Still Uses? appeared first on WEBNET Hosting.

Many thanks for your support and for your donations.

We’ve got quite a bit of news this month.

As usual, the features and improvements presented in this blog post are a preview. They’re still being worked on and are planned for the next version of Linux Mint.

XApp

The XApp project started within Linux Mint but it has grown a lot since then. Its libraries, tools and applications are used in many desktops and distributions.

It was given a new website to make it easier to discover all its components: https://xapp-project.org

Library

Thingy, known as the “Library” in the desktop, was redesigned.

Until now, documents were grouped by application. For instance, PDF files and all the other document types supported by XReader were grouped together under “Document Viewer”.

The library now organizes documents by type instead. It no longer matters which applications are installed or which formats they support.

The sidebar comes with predefined groups for common types of documents. You can choose which groups are visible and change their order.

You can also create your own groups and decide which types of files they contain.

For instance, you could create a group for a particular project, limited to a work directory and grouping multiple document types.

The library also lets you remove individual items from the list or clear the entire list of recently opened documents at once.

Xepub

Xepub is a new EPUB reader.

It provides a reading experience similar to what you would find on a Kindle or a Kobo.

In the desktop it will appear as your “Book Reader”.

You can navigate between chapters, search the contents of the book, add bookmarks and quickly navigate between them.

When you reopen a book, Xepub remembers where you were, so you can simply continue reading.

If you select a word, a menu gives you quick access to search engines, so you can look it up in a dictionary, find it on Wikipedia or translate it.

The search engines are fully configurable in the preferences.

You can change their URLs or add anything you might find useful.

If you select a passage, it gets highlighted as an annotation. You can give it the color of your choice and add a note.

And of course, you can adjust the zoom level, choose your preferred fonts and customize the reading experience to your liking.

Clockenstein

It’s alive! 🙂

Say hello to your “Calendar”.

Clockenstein is a brand new calendar application and the latest addition to the XApp family.

It supports multiple calendars and works with local calendars, Google Calendar and CalDAV services such as Nextcloud.

You can create and edit events and choose which calendar they belong to.

Down in the laboratory, there’s a little more going on. 🙂

When we worked on Clockenstein, one of the things we really wanted was to always know what was going on with remote calendars. Are they connected? Are they synchronized? And if something goes wrong, why?

Clockenstein keeps track of the state of each remote calendar and reports connection and synchronization issues in the application. If a calendar can’t be reached, you can still see its events, but it becomes read-only until the connection is restored.

Clockenstein has its own daemon and its own notification agent. So it doesn’t matter whether the Calendar application is open or not.

The daemon keeps remote calendars synchronized and keeps an eye on upcoming events for calendars that have reminders enabled.

When a reminder is due, the notification agent presents it on the desktop. You can dismiss it or snooze it to be reminded again later.

Finally, Clockenstein provides interfaces for desktop environments to interact with it, and even to replace its notification agent. The Cinnamon clock applet is fully compatible with these interfaces, so just like in the past, you’ll see your calendar events directly from the panel.

Sponsorships:

A total of $15,693 were raised thanks to the generous contributions of 438 donors:

$1000 (3rd donation), Marci S.
$696 (3rd donation), Michael S.
$278 (8th donation),
$250 (4th donation), Chris K.
$232, Lutz D.
$200 (2nd donation), Gary B.
$200, John L.
$174 (26th donation), Naoise G. aka “Gaff”
$155, finn G. P.
$145, adam S.
$139, Simon D.
$139, Valerio R.
$120 (6th donation), George C.
$116 (22nd donation), Jiří B.
$116 (8th donation), Matthias Rainer
$116 (4th donation), Bernard S.
$116, Dirk U.
$116, Lily S.
$116, Markus B.
$106 (2nd donation), Steven M.
$100 (40th donation), Wolfgang P.
$100 (36th donation), John Mc aka “Land Research Project“
$100 (21st donation), James F.
$100 (2nd donation), Mark R.
$100 (2nd donation), Mark S.
$100 (2nd donation), Network Informatics Inc
$100, Alan F.
$100, Anon
$100, Arun S.
$100, CliffordL H.
$100, Parker J.
$100, Patrick S.
$100, Randolph M.
$100, TERRY B.
$100, William B.
$93 (7th donation), Jean-francois R.
$87, Justin J
$81 (7th donation), Flavio Silva Pita Correia
$81, Thomas Z.
$80, Tina H.
$70 (3rd donation), Dainius B.
$58 (27th donation), Bernard H.
$58 (8th donation), Anthony G M.
$58 (7th donation), Hannu H.
$58 (6th donation), António Salsinha – Graphic Designer
$58 (6th donation), Boris I.
$58 (6th donation), Danilo Cesari aka “Dany”
$58 (5th donation), David W.
$58 (5th donation), Derek M.
$58 (4th donation), Pgm R.
$58 (3rd donation), Ronald K.
$58 (2nd donation), Bernhard S.
$58 (2nd donation), Florian W.
$58 (2nd donation), Frank J.
$58 (2nd donation), LAW
$58 (2nd donation), Sebastian K.
$58 (2nd donation), Stichting Elektoor
$58 (2nd donation), Tim N.
$58, A.E.M. J.
$58, Adam K.
$58, Adrian Q.
$58, András L.
$58, Anonymous
$58, Arnaldo P.
$58, Beat B.
$58, Christoph N.
$58, Daniel R.
$58, Dirk H.
$58, Frank H.
$58, Ivars J. aka “neredzams“
$58, Jonas T.
$58, Jorma J. M.
$58, MARC V. D.
$58, Marco K.
$58, mccormack V.
$58, Peter T.
$58, Philip L.
$58, Raymond D.
$58, Rodney S.
$58, Sebastian G.
$58, Telaneo
$58, Tristan P.
$58, Walter E.
$57, Dennis R.
$52 (6th donation), Marius M.
$51, Geoffrey P.
$50 (30th donation), Thomas T. aka “FullTimer1489”
$50 (22nd donation), Mothy
$50 (18th donation), Anthony C. aka “Ciak”
$50 (14th donation), John Eady
$50 (10th donation), Khalid T. aka “k9750”
$50 (8th donation), Michelle and Daniel Grady aka “Pug Masters”
$50 (6th donation), Frank B.
$50 (4th donation), Dr. S. K.
$50 (3rd donation), Henry B.
$50 (3rd donation), Marek N.
$50 (3rd donation), Paul C.
$50 (2nd donation), Marcus H.
$50, Barry K.
$50, Christopher V.
$50, Craig W.
$50, David H.
$50, Edward N.
$50, Eric M.
$50, George D.
$50, George P.
$50, James L.
$50, John aka “isivisi”
$50, John M.
$50, Kye L.
$50, sian M.
$50, Tree B.
$49, aka “bsod”
$46, Daniel D. M.
$44 (4th donation), maurice A.
$40, Robert H.
$40, Robert J. F. J.
$35 (15th donation), H ScottB
$35 (3rd donation), Arttu H.
$35 (3rd donation), Ingo T.
$35 (3rd donation), Jeste S.
$35 (3rd donation), Pauli K.
$35 (2nd donation), Jean P. R.
$35 (2nd donation), Jonathan R.
$35 (2nd donation), Kase E.
$35, arturo C.
$35, Dorian B.
$35, R H.
$34, James M.
$30, Gilbert W.
$30, Jon B.
$30, Justin H.
$30, Matthew T.
$30, Roman P.
$29 (3rd donation), Hans Peter S.
$29 (2nd donation), Mike W.
$29, J.Wolf
$29, Marek K.
$29, Nele S. H. S.
$26 (20th donation), Alessandro S.
$26, Philipp S.
$25 (65th donation), Linux Mint Sverige
$25 (22nd donation), John W.
$25 (18th donation), John N.
$25 (12th donation), Colin S.
$25 (7th donation), David P.
$25 (2nd donation), Daryl R.
$25 (2nd donation), Rick A.
$25, PAULO R. R.
$24 (2nd donation), persson M.
$23 (79th donation), Bryan F.
$23 (73rd donation), Peter E.
$23 (40th donation), Stefan W.
$23 (35th donation), Benjamin W. aka “UncleBens”
$23 (14th donation), Wolfgang N.
$23 (8th donation), Holger B.
$23 (7th donation), Curt S.
$23 (6th donation), Jens R.
$23 (6th donation), Michael L.
$23 (4th donation), Adam B. aka “Neomodus”
$23 (4th donation), Fritz P.
$23 (3rd donation), Antonio S.
$23 (2nd donation), AndererAldi
$23 (2nd donation), Eric S.
$23 (2nd donation), Hartmut H.
$23 (2nd donation), Jens L.
$23 (2nd donation), Krzysztof P.
$23 (2nd donation), Manuel %. Z. B.
$23 (2nd donation), Markus R.
$23 (2nd donation), Matthias H.
$23 (2nd donation), Thomas B.
$23 (2nd donation), Vincent V.
$23, Alexander N.
$23, Ben M.
$23, Benito V.
$23, Christian S.
$23, Christopher L.
$23, Dario G.
$23, Dennis D.
$23, EFTHYMIOS K.
$23, Erich E.
$23, Gerard D.
$23, Heimo K.
$23, Jan B.
$23, Jantelms FSA
$23, Johann M.
$23, John K.
$23, José I.
$23, Jürgen P.
$23, Klaus-Dieter M.
$23, LazyFox.ca aka “A Third Place for the Internet“
$23, Louise M.
$23, Luca A.
$23, Maks B.
$23, Markus F.
$23, Mathilde M.
$23, Matthias S.
$23, Michael M.
$23, Norbert K.
$23, Pasquale P.
$23, Paula S.
$23, Philippe H.
$23, R.V. I. C.
$23, Reinhard H.
$23, Rüdiger K.
$23, Thomas L.
$23, Torbjorn T.
$23, Torsten A.
$23, Ulrich T.
$23, Ursula H.
$22 (2nd donation), Marko H.
$21 (120th donation), Johann J.
$20 (13th donation), Randall W.
$20 (11th donation), Orlando O.
$20 (9th donation), Jerry G.
$20 (7th donation), Edward F.
$20 (7th donation), Oleksandr N.
$20 (6th donation), Juan V.
$20 (4th donation), Perry B.
$20 (4th donation), Theodore K.
$20 (3rd donation), Edward P.
$20 (3rd donation), Jeffrey J.
$20 (3rd donation), Michael S.
$20 (2nd donation), David M.
$20 (2nd donation), Dylan H. aka “D-Guy“
$20 (2nd donation), Justin M. H.
$20 (2nd donation), Kellee C.
$20 (2nd donation), Vincent A. B.
$20, Armen V.
$20, Bill M.
$20, Bruce S.
$20, Carl A. D. F.
$20, Carlos C.
$20, Chris D.
$20, Christopher M.
$20, Clemens K.
$20, Conner M.
$20, Devin G.
$20, Duane L.
$20, Geoffrey C.
$20, George W.
$20, hatem A. N. Y.
$20, Ivan G.
$20, John M.
$20, John P.
$20, Joseph T.
$20, Mark B.
$20, Mark J.
$20, Patrick K.
$20, Peter L.
$20, Philip S.
$20, Ralph F.
$20, Rick L.
$20, Steven H.
$20, Steven T.
$20, Thomas S.
$20, Tomáš %.
$20, William A. C. I.
$20, William M.
$17 (94th donation), Andreas S.
$17, Jevgenijs D.
$17, raoul A.
$17, Vojtch D.
$17, Yosef F.
$16 (2nd donation), Jan H.
$15 (3rd donation), Harris B.
$15, Brian S.
$15, Dushyant S.
$15, Indrapramit B.
$15, Pierce A. C.
$14 (3rd donation), Aleksandre A.
$12 (59th donation), Daniel S.
$12 (51st donation), Tugaleres.com
$12 (32nd donation), Adian K.
$12 (21st donation), Marc V. K.
$12 (19th donation), Abe Z.
$12 (18th donation), Artem Ignatyev aka “ZaZooBred”
$12 (10th donation), Jacques V.
$12 (9th donation), Christoph B.
$12 (7th donation), Ja P.
$12 (7th donation), Stefan B.
$12 (6th donation), Iker L.
$12 (6th donation), Lyuben R.
$12 (5th donation), Bartomiej T.
$12 (5th donation), Thomas G.
$12 (4th donation), Arne A
$12 (4th donation), Jewgeni Smirnow
$12 (3rd donation), Andreas L.
$12 (3rd donation), Heinz S. aka “Heiner“
$12 (3rd donation), Judian G.
$12 (3rd donation), MAH aka “Myne”
$12 (3rd donation), Timothy C.
$12 (3rd donation), Zlatko
$12 (2nd donation), David P. B.
$12 (2nd donation), Koski M. K. aka “Wintageman“
$12 (2nd donation), Mitja S.
$12 (2nd donation), Péter A.
$12 (2nd donation), Thomas H.
$12, Alessandro F.
$12, andre K.
$12, ANDRIS B.
$12, Anton G.
$12, Bartosz W. aka “barthossius”
$12, Bob B.
$12, Bogusław P.
$12, Carina G.
$12, Christopher H.
$12, Daniel G.
$12, Eckhard H.
$12, Egbert N.
$12, Egil B. R.
$12, evangelos K.
$12, Fabian R.
$12, Frank Z.
$12, Giuseppe A.
$12, Günter K.
$12, Helen W.
$12, Henri H.
$12, Hernando C.
$12, Jakub C.
$12, JEROME G.
$12, Johann H.
$12, Jordan L.
$12, Lars-Erik E.
$12, Marc S.
$12, Markus H.
$12, Michal L.
$12, Paweł M.
$12, Philipp W.
$12, Roland B.
$12, Stephan R.
$12, Tamas P.
$12, Tim B.
$12, Tomislav B.
$12, Uwe D.
$12, Wolfgang R.
$12, Wouter V. N.
$12, www.barcelonataxitour.com
$12, Yvan R.
$10 (124th donation), Thomas C.
$10 (66th donation), Philip Woodward
$10 (31st donation), Wilson G.
$10 (15th donation), Mariusz B.
$10 (9th donation), Lars K. aka “laku3008”
$10 (8th donation), Michael G.
$10 (8th donation), Urszula S.
$10 (7th donation), Selwyn A.
$10 (4th donation), kamil P.
$10 (4th donation), Ulf W.
$10 (2nd donation), ANDREW EDMUND JOHN
$10 (2nd donation), David S.
$10 (2nd donation), Giovani D. M.
$10 (2nd donation), Kenneth H.
$10 (2nd donation), Krzysztof aka “O Linuksie Przy Kawie“
$10 (2nd donation), Larry E.
$10 (2nd donation), Victor M. aka “Bull”
$10, Abram M.
$10, Agustin J. C. S.
$10, Amazon Marketing Agency – Vancouver
$10, Bill W.
$10, Clay
$10, Constantinescu V.
$10, Cristian C. P.
$10, David L.
$10, Edgar J. E. C.
$10, Eduardo L. R.
$10, Eliecer C. T.
$10, Hector R.
$10, IGOR Z.
$10, J. Wubben aka “Invirtuate“
$10, John A.
$10, John P.
$10, John R. P.
$10, Leslie C. S.
$10, Lyndsay F.
$10, Michael G.
$10, Michel B.
$10, Narayan T.
$10, Nechaev D.
$10, Noel P.
$10, none
$10, Oleksandr S. aka “electrino”
$10, Oscar C.
$10, Robert K.
$10, Rodrigo S. D. S.
$10, roger B.
$10, sotirios T.
$10, Stephen B.
$10, Three Butterflies
$10, Timothy K. A.
$10, Tyler E.
$10, VANDLE U.
$10, Wesley H.
$9 (2nd donation), SAMUEL K.
$160 from 38 smaller donations

If you want to help Linux Mint with a donation, please visit https://www.linuxmint.com/donors.php

Patrons:

Linux Mint is proudly supported by 2,369 patrons, for a sum of $5,217 per month.

To become a Linux Mint patron, please visit https://www.patreon.com/linux_mint

The post Monthly News – August 2026 appeared first on WEBNET Hosting.

You’ve probably typed pwd countless times without giving it much thought. But when you’re working inside a symlinked directory, writing a shell script that needs the current path, or troubleshooting why a command is running from an unexpected location, understanding how pwd reports the working directory becomes useful.

What pwd Actually Does

pwd stands for print working directory. It shows the absolute path of the directory you’re currently in, starting from the root directory (/).

On most Linux systems, pwd is available in two forms: as a shell builtin in shells such as bash, zsh, and ksh, and as a standalone /bin/pwd command provided by GNU coreutils. These two versions can behave differently when symbolic links are involved.

pwd [OPTION]

The examples in this guide were tested on Ubuntu 26.04 LTS and Fedora 42. The behavior described here is consistent across mainstream Linux distributions using GNU coreutils.

One detail that can easily cause confusion is what happens when you use both -L and -P. If both options are specified, the last option takes effect.

For example, pwd -L -P uses physical mode, while pwd -P -L uses logical mode.

There is also an important difference between the shell builtin and the GNU pwd binary when no option is specified. Bash’s builtin pwd defaults to logical mode (-L), while the GNU coreutils version defaults to physical mode (-P).

This is why the two commands can sometimes show different paths when you are inside a directory reached through a symbolic link.

1. Print Your Current Working Directory

The simplest use of pwd is to print the absolute path of the directory you’re currently in.

pwd
/home/tecmint

2. Create a Symlink and Move Into It

To see how pwd handles symbolic links, create a symlink to a system directory and move into it. We’ll use this directory in the next few examples.

ln -s /var/www/html/ htm
cd htm

3. Print the Logical Path with -L

Once you’re inside the symlinked directory, pwd -L shows the logical path, which preserves the symbolic link you used to reach the directory.

pwd -L

/home/tecmint/htm

4. Print the Physical Path with -P

The -P option shows the physical path instead. It resolves the symbolic link and displays the directory’s actual location on the filesystem.

pwd -P

/var/www/html

5. Check the Default Behavior of /bin/pwd

You can run the standalone pwd binary directly to see how it behaves without any options.

/bin/pwd

/var/www/html

This matches pwd -P, not pwd -L, showing that the GNU coreutils /bin/pwd binary defaults to resolving symbolic links.

Bash’s builtin pwd behaves differently. It tracks the logical path through the $PWD variable and defaults to -L. That’s why pwd and /bin/pwd can sometimes show different paths when you’re working inside a directory reached through a symbolic link.

6. Check Your pwd Version

If you’re troubleshooting a script that relies on a particular coreutils behavior or option, it can help to check which version of pwd you’re using.

/bin/pwd --version

pwd (GNU coreutils) 9.11
pwd (GNU coreutils) 9.11
Copyright (C) 2026 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later .
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.

Common Mistake: Assuming pwd Accepts a Path

Unlike commands such as cd command or ls command, pwd command doesn’t accept a directory path to operate on. It reports the working directory of the current shell.

For example:

pwd /home

Depending on whether you’re running the shell builtin or the standalone binary, you’ll get an error rather than a change of directory. pwd never changes your current directory.

7. Find Every pwd Available on Your System

Your shell may have both a builtin pwd and a standalone coreutils binary. The type command shows which one is being used and where other versions can be found.

type -a pwd

pwd is a shell builtin
pwd is /bin/pwd

This is useful when pwd behaves differently from /bin/pwd, because it lets you confirm exactly which implementation your shell is resolving.

8. Store the Current Directory in a Variable

You can capture the output of pwd in a shell variable, which is useful when writing scripts that need to reuse the current directory.

a=$(pwd)
echo "Current working directory is: $a"

Current working directory is: /home/tecmint

Using command substitution with $(pwd) lets the shell run pwd and assign its output to the variable.

9. Check the Current and Previous Directory

Bash keeps track of both the current directory and the directory you most recently left. These are available through the $PWD and $OLDPWD variables.

echo "$PWD $OLDPWD"

/home /home/tecmint

Here, $PWD contains the current working directory, while $OLDPWD contains the previous one. This is also what makes commands such as cd useful for quickly switching back to the directory you just left.

10. Show the Current Directory in Your Shell Prompt

You can configure Bash’s prompt to show the current directory automatically, so you don’t need to run pwd every time you want to check where you are.

PS1='w$ '

For example:

~/tecmint$

The w in Bash’s PS1 is a prompt escape that expands to the current working directory, using ~ for your home directory. It is preferable to putting $(pwd) directly in PS1, because command substitution in the assignment is evaluated when the assignment runs rather than automatically each time the prompt is displayed.

Common Mistake: Expanding $(pwd) When Setting PS1

It’s easy to assume that PS1="$(pwd)> ” will keep showing your current directory. However, the $(pwd) command is evaluated when you assign the variable, so the prompt can remain stuck at that directory.

Use Bash’s built-in w or W prompt escapes instead which is designed to update automatically whenever the prompt is displayed.

PS1='w$ '

If you genuinely need command substitution in a custom prompt, keep the command inside single quotes so Bash can expand it when the prompt is displayed:

PS1='$(pwd)> '

Then change directories and check the prompt using cd /tmp and the prompt should now reflect the new directory.

11. Find the Absolute Path of the pwd Binary

Useful when more than one version of a command exists across your $PATH.

which pwd

/bin/pwd

Keep in mind that which only finds executables in your $PATH. It won’t identify Bash’s builtin pwd, which is why type -a pwd from the earlier example is more useful when you want to see all available implementations.

12. Find the pwd Manual Page

The location of a command’s manual page can vary between distributions. Instead of assuming a fixed path, let man tell you where the page is installed.

man -w pwd

/usr/share/man/man1/pwd.1.gz

You can then open it normally with:

man pwd

13. Use pwd Inside a Conditional Shell Script

pwd becomes more useful in scripts when you need to check the directory from which a script is running before performing an action. First, create a test directory and script:

mkdir -p ~/tecmint && cd ~/tecmint
nano pwd-check.sh

Add the following:

#!/bin/bash
x="$(pwd)"
if [ "$x" == "/home/$USER/tecmint" ]; then
  echo "You're in the tecmint directory"
else
  echo "Not in tecmint, currently in: $x"
fi

Make the script executable and run it:

chmod +x pwd-check.sh
./pwd-check.sh

You're in the tecmint directory

The script stores the current working directory in x and then compares it with the expected path before continuing.

14. Get the Parent Directory Without Changing Directories

You can combine pwd with dirname to get the parent directory of your current working directory without actually moving there.

cd ~/tecmint/htm 2>/dev/null || cd ~/tecmint
echo "Parent directory: $(dirname "$(pwd)")"

Parent directory: /home/tecmint

Here, pwd supplies the current directory, while dirname removes the final path component to return its parent directory.

15. Compare pwd with realpath

The realpath command resolves the current path to its canonical physical location. Comparing it with pwd -P is a quick way to verify that both commands resolve the same physical path.

cd ~/tecmint/htm
pwd -P
realpath .

/var/www/html
/var/www/html

Both commands resolve the htm symbolic link and show the actual target directory.

Conclusion

By now, you should have a clear idea of how pwd command works beyond simply printing your current directory. You’ve seen how -L and -P handle symbolic links differently, why Bash’s builtin pwd and the /bin/pwd binary can return different paths, and how to use pwd in scripts, shell prompts, and path comparisons.

The easiest way to make this behavior clear is to try it yourself. Find a symlinked directory on your Linux system, cd into it, and run pwd -L followed by pwd -P. Seeing the logical and physical paths side by side makes the difference much easier to understand.

Have you ever run into a script that behaved unexpectedly because of a logical versus physical path, or a shell prompt that showed the wrong directory? Tell us what happened in the comments below.

The post 15 Practical pwd Examples in Linux You Should Know appeared first on WEBNET Hosting.

The post I Built a Fake Home Theater… appeared first on WEBNET Hosting.

The post Creators Give One Word Take To Our New Screwdriver appeared first on WEBNET Hosting.

Summarize with:

Follow:

Buy multiple domain extensions when you need to protect an established brand from squatters, reach customers in specific countries, or cover misspellings and lookalike addresses that could send customers elsewhere. One domain is usually enough when none of these risks applies.

Registering multiple domain names doesn’t improve SEO on its own, so redirect extra domains to your primary website instead of publishing duplicate copies of it.

The right portfolio follows a priority order: primary extension first, then the country-code domain of your main market, then one or two recognized alternatives. If you do need extras, one to three is almost always enough.

Hostinger’s Brand protection page lists every available variation of a name you own and registers the selected ones in a single checkout, so protecting a brand across extensions takes minutes.

When is it recommended to buy multiple domain extensions?

Buying multiple domain extensions is worth it in four situations: protecting your brand identity, targeting customers in specific countries, capturing common misspellings, and preventing customer confusion. Each one removes a concrete risk; when none of them applies, the money is better spent elsewhere.

Brand protection, misspellings, and customer confusion all follow the same logic. Domain registration is first-come, first-served.

Every extension is a separate registration of the same name under a different top-level domain (TLD), a core part ofhow top-level domains work. If you own domain.com, whoever registers domain.net first owns it – whether that’s you, a competitor, or a squatter.

To protect your brand from squatters

Owning the obvious variations of your brand name costs less than recovering them from a stranger. Squatters register versions of known names to resell them at a markup or to trade on their reputation. The most reliable way toprotect your brand from domain squatting is to hold those versions yourself before anyone else does.

A defensive registration runs $10–30 per year. Recovering a name someone else registered means filing a complaint under the Uniform Domain-Name Dispute-Resolution Policy (UDRP). That starts at $1,500 in World Intellectual Property Organization fees alone, before lawyers – and the panel’s decision isn’t guaranteed to go your way. Ten years of renewals cost less than one dispute.

To target customers in specific countries

A country-code top-level domain (ccTLD), like .de, .co.uk, or .com.br, tells local customers they’re buying from someone who operates in their market. Local extensions carry weight where it matters commercially: shoppers are more willing to enter payment details, and local partners read them as a commitment.

Register the ccTLD when you already have customers, local pricing, or fulfillment in that country. Skip it while expansion is still only a plan.

To capture common misspellings

Misspelled versions of your domain send real visitors to whoever owns them. It happens even at the extension level. Someone intending to visit domain.com might type domain.co instead, which is one reason .co can be useful in a defensive domain portfolio.

Name-level typos follow the same logic at a larger scale. Popular brands see misspellings of their name searched hundreds of times a month, and a typo domain in the wrong hands collects that traffic or, worse, imitates the real site. You don’t need to buy all the possible misspellings of your brand, only the one or two that people usually make.

To prevent customer confusion

Customers who guess your address should reach you, not whoever owns the extension they guessed. A competitor who owns the .net domain of your name collects your walk-in traffic, and a lookalike domain in a stranger’s hands provides a ready-made address for phishing emails that impersonate your business.

Paid campaigns, printed materials, and word of mouth all produce visitors who type addresses from memory. Every mistyped guess that sends customers to someone else is a customer you paid to lose.

When is one domain extension enough?

One domain extension is enough when your brand has no traffic worth protecting, your audience is local, and your name is too generic for a squatter to profit from. Squatters target names people already search for, so a new project with no established audience is generally a lower-priority target.

A squatter who registers bestplumbingservices.net can’t resell it to you at a markup, because you could choose from dozens of similar domain names. The risk that justifies a defensive portfolio – someone holding your specific name hostage – doesn’t apply when there are plenty of similar domain names available.

The renewal cost also argues against unnecessary registrations. Five extra domains at $15–30 each cost $75–150 per year, and that figure compounds at renewal, not the first-year promotions registrars advertise.Domain registration and renewal costs vary widely by extension, so a defensive portfolio that looks affordable at registration can look different a year later.

Register extras when the brand earns enough traffic that someone else holding a variation would cost you something real. Before that, the money is better spent on the site itself.

Which domain extensions should you prioritize?

Prioritize domain extensions in this order: your primary extension first, the ccTLD of your main market second, one or two widely recognized alternatives third, and typo variations last. The order follows risk, not price.

Choosing the right primary domain name comes before any defensive buying. Every variation you register inherits whatever name you settle on, so a name that’s easy to mistype means more variations to defend..

Here’s how each type maps to who needs it and when to skip it:

If your preferred primary domain is already registered by someone else, the recognized alternatives in the table are the first place to look – a full breakdown ofalternative extensions when your preferred domain is taken can help narrow the choice.

How many domain extensions do most businesses need?

Most businesses need one primary domain, plus one to three extra extensions. A local business does fine with the primary alone; a national brand adds one or two recognized alternatives, and an international one adds the ccTLD of each market it operates in.

Every registration past the first should have a reason you can name. If you can’t say which risk a domain removes, skip it.

Do multiple domain extensions help SEO?

Multiple domain extensions don’t help SEO on their own. Search engines rank websites and their content, not registration lists, so a domain showing a placeholder page contributes nothing to the site you actually run. Redirecting a freshly registered extension adds nothing either, because a domain nobody has ever linked to has no ranking signals to pass on.

The damage shows up when owners publish on every extension they own: identical sites compete with each other, and search engines pick one version to rank while filtering out the rest.

Country domains are the one partial exception. A ccTLD works as a strong geographic signal, but the benefit arrives with genuinely localized content, such as local language, local currency, and a local address. A .de that redirects to your .com sends no geographic signal at all; it just catches the customers who typed it.

What should you do with your extra domain extensions?

Redirect them. Multiple domains pointing to one website is the point of a defensive portfolio, and a permanent 301 redirect is the mechanism: it forwards visitors and tells search engines which address is the real one.

The steps toredirect additional domains to your main website take minutes per domain with most registrars.

A working portfolio needs four things in place:

1. One primary domain. Content, backlinks, business cards, and ad campaigns all use this single address.
2. A 301 redirect from every extra extension to it. A 302 signals a temporary move, so search engines treat it differently.
3. Valid SSL certificates on the redirected domains. Browsers verify the certificate before following the redirect, so an expired one displays a security warning instead of your homepage.
4. Auto-renewal on every defensive registration. An expired defensive domain returns to the open market, which undoes the point of buying it.

Redirecting also gets confused with transferring, and they solve different problems. Thedifference between redirecting and transferring a domain comes down to what moves: a transfer changes which registrar manages the name, while a redirect changes where visitors end up.

Consolidating a scattered portfolio usually means doing both: transfers to bring every domain under one roof and redirects to point them all to the same website.

How to protect your brand across multiple extensions with Hostinger

Open the Brand protection page in hPanel, select the extensions you want, and register them all in one checkout. The page lists every version of your name with its price, marks the ones you already own with an Owned label, and suggests a ready-made bundle of the most common defensive picks.

Hostinger supports more than 500 extensions, so the page loads every variation of your name at once instead of making youcheck domain availability one at a time. You act on the handful that matter.

Brand protection works on domains already in your Hostinger account. If you’re starting from scratch,buying and registering a domain first takes a few minutes, and once it’s in your account, Brand protection shows every available variation of it.

To secure several variations at once:

1. In hPanel, go to Domains and click Manage next to the desired domain.
2. Select Brand protection in the left sidebar.
3. Click Add to cart on the variations you want, or add the suggested bundle.
4. Click Continue to cart, set the registration term for each domain, and complete the checkout in one payment.

Your contact details apply across the whole cart, and any extension that requires additional information, as some country codes do, gets flagged before payment. Free WHOIS privacy protection is included with each registration, and the cart shows every domain’s renewal rate and renewal date before you commit.

After checkout, each new domain appears in the Domain portfolio with its own Auto-renewal toggle.

How to decide if buying another domain extension is worth it

Another domain extension is worth buying when it protects your brand, supports the market you serve, or costs little relative to the risk it mitigates. Three questions help you check:

• Would a stranger owning this variation cost you customers, revenue, or reputation?
• Do you serve the country behind that ccTLD today?
• Is the renewal price trivial next to the risk it removes?

Register the extension when you can answer yes to the first or second question, and yes to the third. Anything else is a maybe, and maybes belong on a watchlist, not in a cart.

Review the list at renewal time, with the real yearly cost on screen, rather than buying in advance for markets and threats that may never arrive. New extensions launch every year, so this isn’t a decision you make only once.

The brands that handle it well treat their domain list like the rest of their infrastructure: small, current, and tied to a purpose they can name.

All of the tutorial content on this website is subject to

Hostinger’s rigorous editorial standards and values.

The author

Bruno Santana

Bruno is a Content Writer at Hostinger, focused on creating and optimizing helpful, engaging articles about web development and marketing. With a background in journalism, he combines storytelling with practical insights to make complex topics easier to understand. He has also contributed to publications like MacMagazine and Jornal A Tarde. Outside of work, Bruno enjoys exploring art, cooking, and technology.

The post Should you buy multiple domain extensions? appeared first on WEBNET Hosting.

Summarize with:

Follow:

The best Manus alternatives in 2026 fall into three groups: ready-to-use general agents like ChatGPT, specialized business and research tools like Hostinger Agent and Genspark, and open-source options like OpenManus and n8n that you run on your own server.

Each group answers a different Manus limitation: Manus runs every task on its own infrastructure and bills per task in credits rather than a flat rate.

Hosted alternatives solve the pricing problem – you can start in minutes and pay a flat monthly fee. Self-hosted alternatives are free to use and keep your data on your own server, but you handle setup and maintenance yourself. Predictable costs, data control, and your technical skills decide which one fits best.

Best Manus AI alternatives at a glance

The best Manus alternatives cover business operations, research, workflow automation, and software engineering, with starting prices from free to $42.50 per month.

1. Hostinger Agent: role-based agents with guided skills

• Best for: Solopreneurs and small business owners who need specialized agents for SEO, content, marketing, legal, and sales.
• Price: From $6.99/month.
• Open source: No.

Hostinger Agent is Hostinger’s AI assistant for small business owners who manage content, SEO, marketing, legal, and sales without a dedicated team.

It covers the business work that Manus users delegate to a general agent, split across seven domain specialists rather than one assistant you have to prompt and guide. The lineup includes Business Advisor, Creative Writer, SEO Consultant, Marketing Planner, Legal Advisor, Customer Communications, and Sales & Outreach.

Business context carries forward between conversations, so agents already know your brand, preferences, and past decisions each time.

Hostinger Agent key features

• Seven specialized agents. Strategy, writing, SEO, marketing, legal, customer communications, and sales, each with its own skill set rather than one shared prompt.
• 150+ guided skills. SEO audits covering 24 on-page factors, blog posts, logo design, content calendars, pricing reviews, privacy policy drafts, and lead outreach.
• 1,000+ app connections. Acts through Gmail, Slack, WhatsApp, HubSpot, Notion, Jira, GitHub, and Figma without leaving the chat.
• Persistent business memory. Agents keep your brand, preferences, and past decisions across conversations.
• Scheduled recurring tasks. Image and logo generation and file uploads with version tracking.

Hostinger Agent limitations

• No autonomous execution. No independent web browsing, code execution, or sandboxed environment.
• Skill-bounded scope. Requests outside the 150+ built-in workflows lose the guided structure.
• Cloud-hosted only. No self-hosted option, which OpenClaw and OpenManus cover.

Hostinger Agent pricing:

Hostinger Agent starts at $6.99/month on the 12-month plan and renews at $8.99/month, so it is worth factoring that in before committing. Shorter-term plans are available at higher monthly rates for those who prefer more flexibility.

All plans include 1,000 credits per month, refreshing automatically with no manual top-ups needed. Credits cover the full range of tasks across all seven agents: skill runs, image generation, scheduled automations, and file handling. A 30-day money-back guarantee applies to all plans.

2. ChatGPT: general-purpose assistant with built-in browsing

• Best for: Individuals who want a ready-made agent for everyday tasks and browser actions; teams that want persistent Workspace agents.
• Price: Free; individual plans from $8/month; Business from $20/user/month billed annually (minimum 2 seats).
• Open source: No.

ChatGPT is OpenAI’s general-purpose AI agent with built-in browsing, scheduled tasks, and app connections on individual plans, and a separate Workspace agents product for business teams.

For Manus users considering it as a replacement, its agent capabilities are split across two plan types – individual and Business – and only Business includes Workspace agents.

On individual plans (Go, Plus, and Pro), ChatGPT includes a built-in browser on every tier, scheduled tasks from the Go plan upward, and app connections to internal tools on Plus and above. These cover everyday research, automations, and integrations without needing a team account.

Workspace agents require a Business plan with a minimum of two seats. The full product includes a persistent template gallery, an org-wide agent directory, and 24/7 scheduled runs. For solo Manus users, this is the key constraint: the feature that most directly replaces Manus is not available on any individual plan, including Pro from $100/month.

ChatGPT key features

• Model access. GPT-5.6 Sol on Plus, GPT-5.6 Sol Pro on Pro, all frontier models on Business.
• Template gallery and org directory. Workspace agents are shared across the company rather than rebuilt per person.
• 24/7 scheduled runs. Business-tier agents work on a cadence without prompting.

ChatGPT limitations

• Usage limits on every individual plan. Message and feature caps vary by tier and are not always published as fixed numbers.
• Ads on Free and Go. Sponsored placements appear below responses.
• No self-hosted option. Everything runs on OpenAI’s infrastructure, the same constraint that pushes people off Manus.

ChatGPT pricing:

Individual plans run from $0/month (Free) to $8/month (Go), $20/month (Plus), and from $100/month (Pro). Workspace agents require a Business plan at $20/user/month billed annually or $25/user/month on a monthly plan, with a minimum of two seats. Enterprise pricing is custom.

3. Genspark: Super Agent and AI Slides

• Best for: Research-heavy tasks, presentations, and multi-modal content that benefits from parallel AI agents.
• Price: Free (100 credits/day); paid plans from $19.99/month billed annually.
• Open source: No.

Genspark, currently on AI Workspace 6.0, is an AI workspace built around Super Agent and AI Slides, developed for research-heavy work, content creation in text, images, and video, and AI presentations. Super Agent handles complex research tasks autonomously, while GenTeam coordinates multiple AI agents working in parallel.

AI Slides and Deep Research make it the strongest match for readers who use Manus mainly for research and presentations. Super Agent researches the topic first, then builds the output, following the same research-before-execution pattern Manus uses.

The free tier gives 100 credits per day, and agent tasks run on it. Standard requests cost 30 to 150 credits each; Ultra mode, which uses the most capable models, costs 100 to 600 credits per request. A single Ultra task can consume the full daily free allowance.

Genspark key features

• Model choice. Access to multiple major model providers from one workspace.
• Saved workflows. Rerun a research or content process without rebuilding it.
• Commercial content license. Included on paid plans, so generated material is usable in client work.

Genspark limitations

• No public pricing page. Plan details are visible only after you create an account.
• Credits do not carry over. Monthly allocations reset each billing cycle.
• Fair-use throttle. Heavy chat or image use triggers a five-hour cooldown that buying credits will not lift.

Genspark pricing:

Genspark’s free plan gives 100 credits per day, The Plus plan starts at $19.99/month billed annually and includes 10,000 credits per month, 50 GB of storage, unlimited AI Chat and image generation, and a commercial content license.

The Pro plan at $199.99/month raises that to 125,000 credits and 1 TB of storage. Genspark states that the unlimited chat and image perks are guaranteed through December 31, 2026.

4. Lindy: Slack-native AI teammate

• Best for: Teams that want a persistent AI teammate in Slack handling recurring workflows, meetings, and inbox management.
• Price: From $29.99/user/month; 7-day free trial.
• Open source: No.

Lindy is an AI teammate designed to live inside your team’s Slack workspace and handle recurring work, such as inbox management, meeting notes, daily briefs, and scheduled follow-ups. You give it a task, and it does the work.

The credit system maps to the scope of each task. Everyday lookups and replies cost 2 to 250 credits; research tasks and longer reports run 250 to 1,000 credits. Complex dashboard builds and heavy scheduled workflows reach 1,000 to 2,500 credits – close to a full month’s allowance for one seat. Credits pool across the whole team, which means one person’s heavy use draws from everyone’s shared balance.

Lindy also supports computer use, meaning it can operate software directly rather than only connecting through APIs. Combined with MCP support and over 1,000 integrations, it covers most of the recurring operational work Manus users hand off to an agent.

Lindy key features

• Meeting support. Joins calls as a named participant, records, transcribes, and files recordings in shared folders.
• No surprise bills. Lindy pauses and notifies admins when the credit pool runs low, and per-seat caps are configurable.
• 40+ built-in skills. Research, data analysis, decks, and dashboards, plus custom skills your team builds once and reuses.

Lindy limitations

• Slack-centric by design. Teams working elsewhere lose most of the product.
• Per-seat pricing. Costs grow with headcount from the $29.99/user/month base.
• No free tier. Access needs a paid seat once the seven-day trial ends, and direct signups are billed immediately.
• Approval gates. Actions with external consequences stall until a named approver signs off, slowing unattended workflows.

Lindy pricing:

Lindy starts at $29.99/user/month, which includes 3,000 credits per user per month, pooled across the workspace. Higher tiers add more credits and connected inboxes. Enterprise adds HIPAA compliance for healthcare data, company-account login, automatic user syncing, and activity logs.

Credits can be topped up at $10 per 1,000 at any time. The 7-day trial starts automatically when anyone in your Slack first tags Lindy.

5. OpenManus: open-source Manus replication

• Best for: Developers and technical users who want a self-hosted, open-source Manus-style agent with bring-your-own-model flexibility.
• Price: Free (self-hosted; bring your own API keys).
• Open source: Yes (MIT).

OpenManus is an open-source AI agent framework built by researchers from the MetaGPT team as a rebuild of Manus without the credit system. It is one of the earliest and most referenced open-source alternatives in this space, with 58,100+ GitHub stars.

It runs on your machine with your own model API keys and handles the same categories of tasks Manus does: browser automation, file operations, and multi-step agent runs.

Setup requires cloning the repository, adding your model credentials to a config.toml file, and running python main.py. A recent main-branch update integrates Browser Use CLI 3.0, which attaches to a local Chrome session for browser automation without an additional API key.

Two caveats are worth noting before you start: the multi-agent flow is labeled “unstable” in the project’s README, and the latest tagged release, v0.3.0, dates back to April 2025, even though the main branch receives regular commits.

OpenManus key features

• DataAnalysis agent. Optional add-on for data analysis and visualization tasks.
• Local browser mode. Attaches to your existing Chrome session, so browser automation needs no extra API key.
• 59 contributors. Built and maintained by researchers from the MetaGPT team.

OpenManus limitations

• Technical setup required. Python 3.12, dependency installation, and manual configuration, with no graphical interface.
• No managed hosting. You run the infrastructure, cover model API costs, and handle updates.

OpenManus pricing:

OpenManus is free to download and self-host under the MIT license. Running it costs whatever your model provider charges per API call; there is no platform subscription or usage fee.

6. AutoGPT: agent platform with a visual builder

• Best for: Users who want to build and run autonomous agents on a schedule, either in the cloud or self-hosted, with visual workflow control.
• Price: Free (self-hosted); cloud from $42.50/month billed annually.
• Open source: Partially (classic agent is MIT; cloud Platform uses Polyform Shield).

AutoGPT is a platform for building and running autonomous AI agents. It replaces the open-ended runs Manus handles on demand with agents you define once and rerun on a schedule.

The product has four parts. AutoPilot turns a plain-language description into a working agent; Agents is the dashboard for monitoring runs and costs; Marketplace offers ready-made agents you can deploy in one click; and Build is a visual block editor for step-by-step control.

You can run it two ways. The hosted Platform is managed by AutoGPT, requires no model API keys, and uses Polyform Shield, which permits personal and internal business use but prohibits building a competing hosted service on top of it. The classic standalone agent, in the classic/ folder, is MIT-licensed and runs anywhere, on your own infrastructure with your own model keys.

Agent runs on the Platform are pay-as-you-go from a prepaid balance. Your subscription covers AutoPilot chat, while the balance covers whatever your agents consume when they run.

AutoGPT key features

• Browser automation on all plans. Agents browse, scrape, and interact with pages.
• 45+ integrations. Gmail, Google Calendar, GitHub, Slack, HubSpot, Notion, Linear, and Airtable.
• Schedule and trigger support. Run agents on demand, on a schedule, or from a webhook.
• Per-task spending caps. Set a budget per task and watch costs as they accrue.
• Approval steps. Agents pause and ask before taking consequential actions.

AutoGPT limitations

• No free cloud tier. The managed platform needs a paid plan, and self-hosting requires Docker plus your own model keys.
• Platform is in beta. The current release still carries a beta version tag.

AutoGPT pricing:

The self-hosted path is free, with no license fee; you cover infrastructure and model API costs. The cloud Platform starts at $42.50/month, billed annually, for Pro, and at $272/month for Max, which includes 8.5 times Pro’s chat allowance.

The subscription covers AutoPilot chat. The agent runs the bill separately from a prepaid balance and is priced the same on every plan, so a higher tier buys more chat rather than cheaper automation.

7. CrewAI: Python framework for agent crews

• Best for: Developers building production-ready multi-agent systems where different agents have distinct roles and collaborate on complex tasks.
• Price: Free (50 executions/month on CrewAI AMP); enterprise pricing on request.
• Open source: Yes (MIT).

CrewAI is an open-source Python framework for building and coordinatingmulti-agent systems: networks of specialized AI agents that collaborate on complex tasks, each with a defined role, goal, and set of tools. It replaces Manus’ multi-step task handling with agent roles you define in code.

With roughly 29 million monthly downloads from PyPI, Python’s package registry,, it ranks among the most widely usedAI agent builders in the Python ecosystem.

The framework has two core concepts that work together. Crews are autonomous, role-based agent teams built for tasks that benefit from collaborative decision-making, in which agents hand off work to each other as needed. Flows are event-driven workflows that give precise control over the order in which steps run, what data is carried between them, and how the system behaves in production.

Most real-world implementations combine both: Crews for the reasoning, Flows for the structure.

CrewAI AMP Suite is the commercial layer built on top of the framework. It adds a visual Studio editor, runs your agents on CrewAI’s servers with a live view of each step, and lets a person approve consequential actions before they happen. The free tier includes 50 workflow executions per month on CrewAI’s cloud.

CrewAI key features

• Exportable to Python. The visual Studio editor produces code you own rather than a locked configuration.
• Real-time tracing. Every model call, tool call, and memory read is visible while it runs.
• Enterprise controls. SSO, role-based access, PII redaction, and VPC or on-premise deployment.

CrewAI limitations

• Python-first. Building crews takes code, and the Studio editor only covers the simpler configurations.
• Learning curve. Crews and Flows are separate patterns that take time to combine well.

CrewAI pricing:

The open-source framework is free with no restrictions. CrewAI AMP’s Basic plan is also free and includes access to the Studio editor, GitHub integration, and 50 workflow executions per month in CrewAI’s cloud.

Enterprise pricing requires contacting the team and includes a company account login, role-based permissions, automatic removal of personal data, deployment on your own infrastructure, US federal security certification, and a 45-day onboarding program.

8. OpenClaw: local agent inside your chat apps

• Best for: Users who want a private, self-hosted personal agent that runs on their own machine and connects to the chat apps they already use.
• Price: Free (MIT, self-hosted); managed hosting from $7.99/month (annual billing).
• Open source: Yes (MIT).

OpenClaw is an open-source personal AI agent that runs locally on your machine and connects to the messaging apps you already use, including WhatsApp, Telegram, Slack, Discord, iMessage, Signal, and Google Chat.

You set up OpenClaw with a one-line installer, npm, or Docker, then point it at your model provider. It handles the same personal assistant work Manus does, including email, calendar, files, browser sessions, and shell commands, with persistent memory across every conversation. All your data, credentials, and agent state stay on your own infrastructure.

The project has 387,400+ GitHub stars and 2,940 contributors. It is developed under the OpenClaw Foundation, a non-profit organization.

You can sign in with a ChatGPT Plus or Pro subscription instead of a pay-per-token API key, which removes the variable API cost that most self-hosted agent setups carry.

Hostinger runs OpenClaw as a managed service through itsAI automation apps, handling installation, updates, and infrastructure so you keep the privacy benefits of a self-hosted agent without the ongoing upkeep.

OpenClaw key features

• 29 chat channels. WhatsApp, Telegram, Slack, Discord, iMessage, Signal, and Google Chat among them.
• Skills and plugins. Community extensions through ClawHub, and agents can write and install their own.
• Optional sandboxing. Tools run with full host access unless you configure a sandbox.

OpenClaw limitations

• Technical setup required. Connecting model providers, channels, and permissions takes time beyond the install.
• You manage the infrastructure. Updates, uptime, and backups are yours on a self-hosted install.
• Fast-moving project. The release pace has introduced instability, so production installs should track the stable channel.

OpenClaw pricing:

OpenClaw is free to self-host under the MIT license. Your costs depend on which model provider you connect to and how much you use it. A ChatGPT Plus or Pro subscription works as the model source, replacing per-token billing.

Hostinger’s managed setup starts at $7.99/month (annual billing), and it handles the infrastructure for users who want a private agent without the upkeep.

9. n8n: workflow automation with AI nodes

• Best for: Technical teams who want to build structured AI workflows they can inspect, version, and control, without per-seat pricing.
• Price: Free (Community Edition, self-hosted); cloud from $20/month billed annually.
• Open source: No (fair-code; free to self-host under the Sustainable Use License).

n8n is a workflow automation platform that combines a visual node editor, AI capabilities, and flexible self-hosting. It sits at the intersection ofAI agents and automation: not an autonomous agent that makes its own decisions, but a programmable system where you design the process and AI handles the steps you assign it.

The platform supportsagentic workflows, retrieval from your own documents (RAG), and connections to outside tools through MCP, with every reasoning step visible on the canvas.. You build in the UI, drop into JavaScript or Python when needed, and replay individual steps without re-running the whole workflow. Testing against real data before you deploy is built in.

It replaces the Manus tasks you run repeatedly, trading autonomous decisions for a process you define once. The pricing model reflects that: you pay per complete workflow execution, regardless of how many nodes the workflow contains.

The Starter cloud plan includes unlimited users, making n8n competitive for teams that run into per-seat costs elsewhere. Self-hosting removes per-seat cost entirely, and managed n8n hosting covers the Docker setup, updates, and backups.

n8n key features

• 400+ integrations. Native nodes for common apps, plus HTTP request nodes for anything without one.
• Step-level replay. Rerun a single step instead of the whole workflow while debugging.
• Git-based version control. Available from the Business tier, alongside environments and audit logging.

n8n limitations

• Fair-code license. Internal business use is allowed, but this is not an OSI-approved license and reselling n8n as a hosted service is restricted.
• AI Assistant is in preview. The built-in workflow-building assistant is still marked preview on cloud plans.

n8n pricing:

The Community Edition is free to self-host with no user limit, no execution caps, and no license fee. Cloud plans start at $20/month billed annually (Starter: 2,500 workflow executions per month, unlimited users, 2,300 AI credits).

The Pro plan at $50/month raises the execution limit to 10,000 and adds workflow history, execution search, and admin roles. Business and Enterprise tiers add Git-based version control, environments, SSO, and audit logging.

10. Devin: AI software engineer

• Best for: Software engineering teams that want an autonomous agent for coding, code review, testing, incident triage, and codebase documentation.
• Price: Free ($0, light quota); Pro $20/month; Max $200/month.
• Open source: No.

Devin is an AI software engineer built by Cognition AI. It handles the coding work that typically requires a human developer: writing and reviewing code, fixing bugs, triaging incidents, and shipping pull requests.

It replaces the coding work Manus handles, and it is built specifically for engineering teams rather than general-purpose use.

The platform has four parts: Devin Cloud for autonomous agent runs in a managed environment; Devin Desktop for inline completions and edits in your code editor; Devin CLI for running Devin from the terminal; and Devin Review for automated pull request review and visual checks on the interface.

DeepWiki, Cognition’s companion codebase documentation tool, auto-generates documentation and system diagrams for existing codebases. Fleet mode lets you run agents in parallel across large migrations and multi-repo projects.

For the open-source equivalent, OpenHands (formerly OpenDevin) covers similar ground under a permissive license, and is worth exploring if self-hosting matters more than the managed experience.

Devin key features

• Full development environment. Browser, terminal, code editor, and test runner, with Devin responding to failed builds and review comments on its own.
• Playbooks. Reusable procedure files that tell Devin how your team handles a task, so the same steps repeat across sessions instead of being re-explained each time.
• Integrations. GitHub, GitLab, Bitbucket, Slack, Teams, Linear, and Jira on Pro and above.

Devin limitations

• Free tier cannot run cloud agents. It covers a light quota and a restricted model set, with Devin Cloud starting at Pro.
• Concurrent session cap. Pro and Max allow 10 at once, and unlimited sessions require the Team plan.
• Team pricing is additive. $80/month base plus $40/month per developer seat.

Devin pricing:

The free plan includes a light agent quota, unlimited inline edits, and unlimited Tab completions, which covers testing and occasional use. Pro costs $20/month and adds access to the latest models, Devin Cloud agents, and up to 10 concurrent sessions. Max costs $200/month, and raises every usage allowance, with the same 10 concurrent session cap.

Teams start at $80/month plus $40/month per full developer seat, with unlimited team members and concurrent sessions. Usage allowances refresh daily and weekly; cost per session varies by model and task complexity.

Why look for a Manus alternative?

Three characteristics separate Manus from most alternatives here: credit-based pricing, cloud-only execution, and an ownership structure that reversed in April 2026.

Pro starts at $20/month for 4,000 monthly credits, and the next tier costs $40/month for 8,000, both with 20 concurrent tasks. The rate per credit is the same either way, so upgrading buys volume, not a better price.

The free plan runs only Manus 1.6 Lite in Agent Mode, with 300 credits refreshing daily and one concurrent task. The daily allowance looks generous, but Manus scopes those credits to “quickly answering simple questions,” and they don’t accumulate, so real agent work still needs a paid tier.

Meta closed its roughly $2 billion acquisition of Manus on December 29, 2025. According to TechCrunch, China’s National Development and Reform Commission ordered both parties to withdraw the transaction on April 27, 2026.

Manus announced its return to independent operations on August 11, 2026. Accounts in some jurisdictions had data created on or after December 29, 2025, deleted on August 23 and 24, with a backup window that closed the morning of August 23.

For anyone relying on the platform for ongoing or sensitive work, that kind of disruption is hard to plan around.

Which Manus alternative should you choose?

The right pick depends on what you were actually using Manus for.

For general autonomous tasks: ChatGPT. Built-in browsing and scheduled tasks on individual plans from $8/month, with no credit system to track. AutoGPT if you want the same open-ended runs on your own infrastructure.

For research and presentations: Genspark. Includes a free tier included, Super Agent follows the same research-before-execution pattern as Manus, and AI Slides covers the deck use case directly.

For business operations: Hostinger Agent. Guided specialists for content, SEO, marketing, legal, and sales, for less than Manus’ cheapest paid plan.

For software engineering: Devin. Purpose-built for the full development lifecycle: code review, incident triage, and PR shipping. Free plan available.

For team workflows on a schedule: Lindy, if your team is in Slack and you want an AI that acts like a colleague; n8n if you need to define every step and inspect every run. Both let you cap spend before a run.

For privacy and control over your own machine: OpenClaw. The most active self-hosted option, with one of the largest open-source contributor communities. Managed hosting is available through Hostinger if you want the privacy benefits without maintaining a server.

For building at the framework level: OpenManus for a standalone self-hosted agent; CrewAI if you need to orchestrate multiple specialized agents in production code.

If your main Manus use case was building web or mobile apps, that’s a different category: anAI website builder will serve you better than any agent here.

Spending caps and approval gates were niche features a year ago. Lindy, AutoGPT, and n8n now ship them by default, and that is the first thing to check on whatever you pick.

All of the tutorial content on this website is subject to

Hostinger’s rigorous editorial standards and values.

The author

Bruno Santana

Bruno is a Content Writer at Hostinger, focused on creating and optimizing helpful, engaging articles about web development and marketing. With a background in journalism, he combines storytelling with practical insights to make complex topics easier to understand. He has also contributed to publications like MacMagazine and Jornal A Tarde. Outside of work, Bruno enjoys exploring art, cooking, and technology.

The post 10 best Manus alternatives in 2026 appeared first on WEBNET Hosting.

It’s easy to install an AI agent, run it with all the guardrails turned off, and only check the sandbox settings after it has changed a configuration file you needed. Codex provides three sandbox levels along with a separate approval policy. The combination you choose determines what it can access and modify.

So before getting into the useful stuff, we’ll install Codex, sign in, and then look closely at the flags that control what it can and cannot touch.

What Codex Does on a Linux Machine

Codex CLI is a local agent that works directly in your terminal. It can read your repository, suggest and make edits, run ls command, grep command, execute your test suite, and show you a diff of the changes it made.

For a sysadmin, that last part is especially important. Codex isn’t just a chat window that prints code for you to copy and paste; it can actually run commands in your shell and make changes on your system. That’s why the sandbox section below is worth reading carefully.

The examples in this guide were tested on Ubuntu 26.04 and RHEL 10. The commands should also work on any modern Linux distribution running on a 64-bit x86 or ARM system.

Install Codex CLI on Linux

The quickest way to install Codex CLI is with the official install script. It downloads a prebuilt binary and places it on your PATH.

curl -fsSL https://chatgpt.com/codex/install.sh | sh

Here’s what each part of the command does:

• curl downloads the install script over HTTPS.
• -f makes curl fail on server errors instead of saving an HTML error page.
• -s runs silently, while -S shows real error messages.
• -L follows redirects used by the download endpoint.
• | sh sends the downloaded script directly to the shell for execution.

That last part comes with an important warning. Piping a remote script directly into sh means the shell executes whatever the server sends. If you’re installing Codex on a system you don’t own or manage, download the script first, review it, and then run it:

curl -fsSL https://chatgpt.com/codex/install.sh -o codex-install.sh

Then inspect the script before executing it.

If you prefer using Node.js, install Node.js and npm first. The package manager command depends on your distribution, so use the one that matches your system.

On Ubuntu/Debian:

sudo apt install nodejs npm

On RHEL/Rocky Linux:

sudo dnf install nodejs npm

sudo runs the command with root privileges. You need it here because the package managers may need to write to system directories. Without it, you may see a Permission denied error.

Once Node.js and npm are installed, install Codex CLI globally:

npm install -g @openai/codex

There’s also a third option if you don’t want to use a package manager; the latest GitHub release provides static binaries. On a 64-bit Intel or AMD system, download codex-x86_64-unknown-linux-musl.tar.gz, extract it, rename the binary to codex, and move it to a directory in your PATH.

Whichever installation method you use, verify that Codex is available:

codex --version

If you see a version number instead of command not found, Codex is installed and available on your PATH. If it isn’t, check the troubleshooting section near the end of the guide. In most cases, the problem is simply that the directory containing the binary isn’t in your PATH.

Sign In From a Headless Server

The first time you run codex, it asks you to sign in. On a desktop, Codex opens your browser and completes the ChatGPT OAuth sign-in automatically.

On a server you connect to over SSH, there’s usually no browser available to open. This is where the device authentication flow comes in:

codex login --device-auth

Codex displays a short code and a URL. Open the URL on your laptop, enter the code, and the server completes authentication without needing to launch a browser locally.

Codex is included with ChatGPT Plus, Pro, Business, Edu, and Enterprise plans, so signing in with your ChatGPT account is the usual approach. If you’re using Codex in automation instead, you can provide an API key through an environment variable:

printenv OPENAI_API_KEY | codex login --with-api-key

You can check your authentication status at any time:

codex login status

This shows the active authentication method and exits with status 0 when valid credentials are available, making it useful as a simple check at the beginning of a shell script.

Sandbox Modes: The Setting That Actually Matters

Codex separates two questions that are easy to mix up: What can the agent access? And when should it stop and ask you?

The sandbox controls the first question with --sandbox, or -s for short:

• read-only lets Codex read files and run commands that don’t modify anything. Nothing on disk is changed.
• workspace-write lets Codex modify files inside the current working directory, but not elsewhere.
• danger-full-access removes the sandbox boundary completely and allows commands to access anything your user account can access.

The approval policy controls the second question with --ask-for-approval, or -a:

• untrusted pauses for commands Codex doesn’t already consider safe.
• on-request lets Codex ask when it needs to perform an action outside the sandbox.
• never runs the entire session without asking for approval.

For everyday work in a repository, the combination worth remembering is:

codex --sandbox workspace-write --ask-for-approval on-request

This gives Codex enough access to edit code and run tests without asking for approval every few seconds, while still keeping it inside the directory where you started it. If it needs to access something outside that directory, you get a chance to approve it.

If the task genuinely requires access to another directory, grant access to that directory explicitly instead of weakening the sandbox:

codex --sandbox workspace-write --add-dir /var/www/

Angle brackets such as <your-site> are placeholders used throughout this article. Replace them with your actual value and remove the brackets.

The Flag You Shouldn’t Reach For

Codex also has a --dangerously-bypass-approvals-and-sandbox flag, aliased as --yolo, which disables both approval prompts and sandboxing. The name is a pretty good warning.

The safest place to use this flag is inside a container or a throwaway VM that is already isolated at the operating-system level. On a machine containing real data, workspace-write with on-request gives you a good balance of speed and protection.

Slash Commands Worth Knowing

Once you’re inside the terminal UI, type / to open the command popup. These are the commands you’ll use most often:

• /status shows the active model, approval policy, writable roots, and how much context you have left.
• /diff shows the Git diff of everything Codex changed, including files that Git isn’t tracking yet.
• /review asks Codex to review your working tree and look for behavior changes or missing tests.
• /compact summarizes the conversation so far and frees up context during a long session.
• /permissions changes the approval preset without restarting Codex.
• /model switches the active model before you start a more demanding task.

Run /status as soon as you start a session. It’s the quickest way to confirm that Codex is using the sandbox you expect. It can also help catch situations where an existing ~/.codex/config.toml setting is overriding the flags you passed on the command line.

Running Codex Against a Local Model

If you want to keep your code and requests on your own hardware, --oss lets Codex use a local model provider instead of OpenAI’s hosted models:

codex --oss --local-provider ollama

This sends requests to Ollama running on the same machine. LM Studio is also supported with --local-provider lmstudio:

codex --oss --local-provider lmstudio

The trade-off is performance and capability. Local models may not perform as well as hosted models, especially on larger or more complex coding tasks, but they give you more control over where your code and requests are processed.

Run Codex Non-Interactively With codex exec

The terminal UI works well when you’re sitting at the keyboard. For cron jobs, CI pipelines, and other tasks that need to run without a person watching, use codex exec, which is also available as codex e.

codex exec --sandbox read-only -o /tmp/audit.md "Summarise every systemd unit in this repo that runs as root"

Here’s what each part does:

• codex exec runs a single task and exits instead of opening the terminal UI.
• --sandbox read-only prevents the run from making changes, which is what you want for an audit.
• -o /tmp/audit.md saves the final response to a file that you can review, email, or commit.
• The quoted string is the prompt. If you use - instead, Codex reads the prompt from standard input.

Two more options become useful when you put Codex into scripts. --json outputs newline-delimited JSON events instead of formatted text, making it easy to pipe the output into jq and use the results in your script. --skip-git-repo-check allows Codex to run in a directory that isn’t a Git repository, which it otherwise refuses to use.

If a scripted run stops partway through and you want to continue instead of starting again, use:

codex exec resume --last

This resumes the most recent session from the current directory.

Tell Codex Your Rules With AGENTS.md

Codex reads an AGENTS.md file from the directory where it is working and uses it as a set of standing instructions. You can create a starter file from inside a Codex session with:

/init

This creates a basic scaffold in the current directory. Edit it to include the rules that matter for your project, such as which test command to run, which directories are generated and should not be edited manually, and any deployment step that Codex must never run automatically.

Once you’re happy with the instructions, commit the AGENTS.md file to your repository. That way, you don’t have to explain the same project conventions at the start of every session.

When Codex Won’t Start

The most common problem immediately after installation is a shell that can’t find the codex command.

codex: command not found

Both the install script and npm install -g place the codex binary in a directory that your current shell may not have picked up yet. Start by opening a new shell and try again. If the command still isn’t found, check where npm installs global binaries:

npm config get prefix

Then add the corresponding bin directory to your PATH in ~/.bashrc. Reload the shell and run:

codex --version

If it prints the version, your PATH is configured correctly.

Everything Else

Codex includes its own diagnostic command, which checks your installation, configuration, authentication, runtime, Git, and terminal setup in one pass:

codex doctor --summary

The --summary flag shows grouped check results and a final count instead of the full diagnostic report. That’s usually enough to quickly identify what’s wrong. When you need more detail, run the command without --summary. If you’re attaching the output to a bug report, add –-json; the JSON output is redacted.

There’s also a version-related detail worth knowing. The old --full-auto flag is deprecated and displays a warning, so use --sandbox workspace-write instead. Likewise, codex mcp-server has been replaced by the Codex app server.

Conclusion

You’ve now installed codex using a script, npm package, or static binary, signed in from a headless server with the device code flow, and learned how --sandbox and --ask-for-approval control what the agent can access and when it needs your approval.

You’ve also seen how codex exec lets you run tasks without the interactive terminal UI, making it useful for scripts and automation. Here’s a good way to start. Pick a repository you know well, run:

codex --sandbox read-only

Then ask Codex to explain a file you’re already familiar with. Because the session is read-only, it can’t modify anything. You can evaluate the quality of its answers before giving it permission to make changes.

What’s the first task you’d trust Codex with on a real server, and what’s the one you wouldn’t? Share it in the comments. The boundaries people set around these tools can be just as interesting as the tools themselves.

The post Codex CLI: Use OpenAI’s AI Coding Agent in the Linux Terminal appeared first on WEBNET Hosting.