New & Emerging

Linux File Permissions

Last updated 2026-09-22

What's new

2026-09-22
  • Codex (a tool by OpenAI that helps you do tasks with AI, like writing, designing, or coding) can be used to build skills, create branded deliverables, and even automate tasks, all without needing a technical background.
  • The Codex desktop app (a program you download to use Codex easily) is recommended for a consistent experience, and it uses the same subscription as ChatGPT (a popular AI chatbot), so you won't need a new account.
  • Codex is more powerful than Work (a tool for non-technical knowledge work) and can do everything Work can do, plus more, making it a better investment for learning and using in the long run.
  • The course will teach you how to use Codex effectively with natural language (regular English, not code) and explain core concepts simply, helping you become a pro AI builder.
2026-08-10
  • This update introduces Cloud Code, a tool (software you pay for monthly online) that helps automate marketing tasks, like creating ads, personalizing emails, and setting up appointment systems.
  • The course teaches how to use Cloud Code at different levels, from simple prompts to advanced cloud routines (automated tasks that run without your input).
  • You'll learn to build your own analytics platform (a system to collect and track data) and automate follow-ups, making your marketing efforts more efficient.
  • Cloud Code requires a paid subscription, but the investment can lead to significant returns, as demonstrated by the instructor's business success.

Key points

What it is

  • Linux file permissions (rules controlling who can read, write, or run files) determine access for the owner, group, and everyone else.
  • They appear as three groups of letters (`r`, `w`, `x`) when you run `ls -l` in the terminal, with `x` meaning "run" for files and "enter" for directories.
  • The Linux file system is a single, unified structure starting at the root directory (the forward slash), unlike Windows with separate drives.

How to use it

  • Start by running `ls -l` to see file permissions and `pwd` to check your current location in the file system.
  • When you get a "permission denied" error, ask what exact file needs access, why it needs it, and if you can fix the problem by changing the project setup.
  • Use an "inspection-first rule" to understand what a device or system file does before permitting writes, and start in a cautious mode that asks before edits.

Watch out for

  • Never use a recursive permission change (changing everything in a folder at once) casually, and avoid giving AI coding agents (software that automates coding tasks) too broad permissions.
  • Be cautious of tools that offer a "dangerously skip permissions" mode, as it bypasses all approval prompts and poses a high risk to your files.
  • Always ensure version control (tracking file changes over time) by having humans, agents, and backups each maintain separate tiers of the file structure.

Lesson 1: What is Linux File Permissions and why it matters

Linux file permissions control who can read, write, or run a file or directory. Run `ls -l` to see them—they appear as three groups of letters: owner, group, and everyone else. The letter `r` lets someone view contents, `w` lets them change contents, and `x` lets them run a file. For directories, `x` means enter or traverse. This matters for AI development because AI coding agents create, edit, and delete files on your system. If an agent runs with too broad permissions (rules granting access), it could harm original files or modify critical system paths. That’s why you give agents a copy of your files to work in, not the originals, and start in approval mode (AI asking before each action). Match access to risk—read-only for diagnostics, write access only for the specific folders you choose. As the agent proves reliable, you can grant more freedom, but excessive permissions (overly broad rules) are a common danger, especially in reusable skills where one person’s setup might let an AI modify many files unexpectedly. Always inspect what an agent can access before allowing writes.

Sources

Lesson 2: How to use Linux File Permissions: step-by-step

Linux file permissions control who can read, write, or run files. Start by running `ls -l` in your terminal. The output shows ownership and permission symbols grouped into three audiences: the owner, the group, and everyone else. The letter `r` allows viewing contents, `w` allows changing contents, and `x` allows running a file. For directories, `x` means something different—it lets someone enter or traverse that directory.

The entire file system starts at the root directory (the forward slash). Unlike Windows with separate drives, Linux uses one unified tree. Your internal disk, external drives, network shares, and even virtual system information all appear inside this single structure. Run `pwd` to see your current location.

When you get a "permission denied" error, don't grab administrator access reflexively. Ask three questions: What exact file does it need? Why does it need elevated access? Can you fix the problem by changing the project setup instead? For example, a `deploy.sh` script failing likely needs one specific permission fix, not full system access. Never use a recursive permission change (changing everything in a folder at once) casually. Fix the specific access problem.

A coding agent should start in the project directory—that's where it inspects and changes files. System directories, service keys, and private SSH files should remain protected. System logs may be useful for troubleshooting but can contain sensitive data. The best workflow is clear boundaries, visible changes, and reversible actions, not unlimited access.

Sources

Lesson 3: Best practices and pitfalls

Linux file permissions (rules controlling who can read, write, or run files) are a common source of "permission denied" errors, especially when using AI coding agents. The commands `ls -l` shows ownership and permissions for three audiences: owner, group, and everyone else. The letter `r` allows viewing contents, `w` allows changing them, and `x` lets you run a file. For directories, `x` means you can enter or traverse (move through) them—a key distinction.

When an AI agent hits a permission error, it is likely trying to write where it lacks `w` access. Many tools offer a "dangerously skip permissions" mode that bypasses all approval prompts. One creator notes this is "complete freedom, but it's at high risk to your files." Another explains that while this speeds up work, it is called "dangerous" for a reason. The smarter approach is to explicitly allow safe commands and explicitly deny destructive ones, like deletes.

Best practice is to start in a cautious mode that asks before edits. Some tools have an "auto mode" that approves routine actions but still protects critical files. For AI agents, use an "inspection-first rule" — understand what a device or system file does before permitting writes. Diagnostic paths (interface points) are not invitations to modify the kernel or hardware.

Finally, remember the Linux file system is not a random collection of folders; it's a structured map for software, configuration, logs, and devices. Knowing where files belong helps you set proper permissions and give agents safe boundaries. Always ensure version control (tracking file changes over time) by having humans, agents, and backups each maintain separate tiers of the file structure.

Sources