Add a token count linter

[aatuvai]

Context

AI agents can process a limited amount of tokens at a time: this limit is called a context window. In order for an agent to keep all relevant information in its memory, the amount of tokens it reads should be manageable.

AI agents are actively reading the Teleport documentation and completing given tasks with the help of the docs. To prevent agents from filling their context window prematurely, overly long documentation pages should be prevented.

Summary

This PR adds a linter to ensure that the token count of any specific docs page does not surpass a pre-defined token count limit. Initially this limit is set to 30,000 tokens (the source for the amount is in this article)

Changes

• Add the server/lint-token-count linter.
• Add tests for the linter in server/lint-token-count.test.ts
• Enable the linter in .remarkrc.mjs

[github-actions]

Amplify deployment status

Branch	Commit	Job ID	Status	Preview	Updated (UTC)
aatuvai/2026-07-01-token-count-linter	b2928a2	4	✅SUCCEED	aatuvai-2026-07-01-token-count-linter	2026-07-01 11:35:52

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: b2928a23c1

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[ptgott]

Thanks! We'll need to fix the violations before we can merge this. There are two:

• docs/pages/changelog.mdx
• docs/pages/reference/audit-events.mdx

We could split up the audit event reference by editing the generator per this issue. We may need to ignore the changelog since it's closely coupled to engineering org practices.