About CodeNx
CodeNx publishes short, tested engineering guides written by people who build and operate real systems. Every article states the exact versions it was verified against, cites primary sources, and shows the commands and expected output.
- Published guides
- 28
- Topics covered
- 25
- Authors
- 2
What CodeNx publishes
CodeNx covers the parts of software delivery that are easy to get wrong and badly documented: running your own infrastructure, deploying to the edge, wiring up automation, and using AI tooling inside real codebases rather than in demos.
Articles are short on purpose. A quick fix is 400–800 words. A tutorial is rarely over 1,800. Length is never padded to hit a word count, and there is no introduction explaining why the topic matters before the instructions start.
How articles are written
-
Every guide states its versions
Articles carry a `testedWith` block naming the exact operating system, runtime and tool versions the instructions were run against, plus the date they were last verified.
-
The answer comes first
Each article opens with a direct answer. If that is all you needed, you can stop reading. Background comes after the solution, not before it.
-
Primary sources only
Citations point at official documentation, source repositories, standards and release notes. Where something is uncertain or version-sensitive, the article says so instead of asserting it.
-
Production reality included
Guides cover the failure modes, the security implications and what to do when the happy path does not happen — because that is where most of the time actually goes.
Where AI fits in
Some articles here are drafted with AI assistance. When that is the case, the article says so at the top — there is no hidden AI content.
An AI-assisted draft is not a published article. Every draft goes through a
verification pass that checks commands, package names, ports, file paths,
configuration syntax and source URLs, followed by a human edit. Nothing reaches
the site while humanReviewed is false, and the publishing pipeline
refuses to build an article that has not been reviewed.
Where a claim could not be verified, the article marks it as unverified rather than presenting it as fact. Fabricated sources, invented benchmarks and claimed first-hand experience that did not happen are treated as defects, not stylistic choices.
Corrections
Software moves; guides go stale. Articles that have not been re-verified within the freshness window display a notice saying when they were last checked.
If something here is wrong, out of date, or no longer works,
tell us — corrections are made in the open and the
lastVerifiedAt date is updated when the steps are re-run.