How statuslin.es checks mods and status lines

Every status line and mod passes these four checks before you can install it, and every update passes them again.

Pinned to one version

A mod is pinned to one commit of its GitHub repository, and a status line to the SHA-256 hash of its script. Installing gets you that version, whatever the author pushes later.

{
  "name": "narrator",
  "source": {
    "source": "git-subdir",
    "url": "https://github.com/midudev/better-claude.git",
    "path": "plugins/narrator",
    "sha": "a2babd9fdc2470ab7541e993b3e6cce55c46d032"
  }
}

Narrator's entry in statuslin.es/marketplace.json, the file Claude Code installs mods from. Its sha is the pinned commit.

Run in a sandbox

Before it's listed, it runs in a fresh sandbox that records the previews on its page. A status line runs in 8 sample sessions, and a mod in the terminal and in Claude Desktop. The sandbox can't reach the internet unless the hosts it needs are approved, and it's thrown away after each run.

Clean repo


  Opus 4.8 high ✦ │ ctx 22% │ app (main) │ PR #1287 ● │ $0.41 · 10m +128 -34
  Session ● ● ● ● ● ● ● ● ○ ○ ○  74% left  Resets in 2h 6m
  Weekly  ● ● ● ● ● ● ● ● ● ● ○  93% left  Resets in 2d 0h

New session


  Opus 4.8 high │ ctx 0% │ app (main) │ $0.00 · 10m

Dirty branch


  Sonnet 4.6 medium ✦ │ ctx 48% │ app ⎇ feat-auth* │ PR #1290 ● │ $0.41 · 10m +128 -34
  Session ● ● ● ● ● ● ● ○ ○ ○ ○  60% left  Resets in 1h 11m
  Weekly  ● ● ● ● ● ● ● ● ● ○ ○  82% left  Resets in 2d 21h

Adaptive Cost Panel, as the sandbox recorded it.

The sandbox is only for previews

Installed, a status line or mod runs on your machine with your permissions.

Reviewed before it's listed

When a status line is submitted, checks look for hidden or encoded code and for reads of credentials such as SSH or AWS keys, and refuse it if they find either. Nothing goes live on its own. Review looks at:

Its source, at the pinned version
The previews the sandbox recorded
The hosts it asks to connect to, and whether it reads your Claude login
For a mod, the events it hooks and the $ calls it makes, from claude plugin validate

An update is a new version and goes through every step again. The live version stays up until the update is approved.

Listed with what it can reach

Its page shows what it can reach, right by the install button. For a mod, that comes from its footprint. For a status line, it comes from the checks on its script and the hosts approved for it. None of it comes from the author's description.

What this mod can do

Reviewed at commit a2babd9
Runs programs
Sends its own requests to Claude, which uses your tokens
Reads environment variables
Sees each prompt you send and can change or drop it
Sees each tool call before it runs and can block or change it
Also uses $.audio.speak, which this page doesn't describe yet

4 more things it can do

Remembers data between sessions
Adds the /narrator command
Adds the /recap command
Runs on a timer

Good to know (3)

Each narrated turn makes a Haiku call, which uses your tokens.
Voices: Siri's on macOS, Narrator's on Windows, speech-dispatcher or espeak on Linux.
It stays on in later sessions until you switch it off.

Are mods safe?

Narrator's panel, as its own page shows it.

If something turns up after it's listed, the listing can come down at once.

All of it is open source

statuslin.es is open source. The sandbox, the checks on submitted scripts, the review queue and /marketplace.json are code in its public repository, so you can read exactly what each step does.

Read it at statuslin.es on GitHub. If you find a way around a check, SECURITY.md says how to report it.

What review can't promise

Review lowers the risk. It can't rule it out.
A reviewer can miss something.
Read the source from its page before you install, and again before you update.

Are Claude Code mods safe? covers what a mod can do once installed and how to check one yourself.