Measuring Go Landlock Adoption 2026

I found a new way to burn tokens

If you maintain a software library, it is easy now to crawl code repositories for your users in Open Source and get statistics on who is using it.

šŸ‘‰šŸ» Read the full statistics with interactive breakdowns and visualizations šŸ’”

Landlock users in Go, by library

These are Landlock-using programs in Go, broken down by the library that they are using to access it:

First of all, I had a hunch that I saw more vibe-coded sandboxing projects in 2026, but had not expected it to be that pronounced. o_O

Some insights:

A. Most programs use go-landlock, which is the official Landlock library for Go that I maintain. 🩵

B. A growing number of programs (34) are using raw system calls. Based on spot checks, it seems that these are mostly vibe-coded, and I suspect that their authors may be unaware.

Not using Go-Landlock yet?

A known valid problem with Go-Landlock is that it’s not possible to start a landlocked subprocess without also landlocking the parent. This is a limitation in Go which we are addressing (but it will require the upcoming Go 1.28 and only be fixed in early 2027).

C. The rest is the long tail:

Landlock feature sets in use

We also have great statistics now about the Landlock ABI level (sandboxing feature set) at which users are using the Go-Landlock library:

9 ABIĀ v1 3 ABIĀ v2 16 ABIĀ v3 13 ABIĀ v4 30 ABIĀ v5 3 ABIĀ v6 5 ABIĀ v7 2 ABIĀ v8 11 ABIĀ v9 6 ABIĀ v10 4 custom

(When reading these graphs, it must be taken into account that new ABI versions were not introduced at an even pace, but it’s anyway a good indicator where there is opportunity for improvement.)

In an ideal world, everyone would always upgrade to use the highest available Landlock ABI version, but it’s of course clear that this is not always possible.

How to use more Landlock features

To upgrade your dependency on Go-Landlock, use a prompt like this (or equivalent):

Upgrade the dependency on go-landlock
and use the upgrading instructions that ship with the library.

The upgrading instructions aim to orient agents and humans alike.

The agent should then attempt to use Landlock at a higher ABI level (with stronger constraints on the sandboxed program). It will read the documented constraints and documented common pitfalls and will try to assist you in ensuring that the Landlock policy stays compatible with your application logic.

Comments