feat: add llms markdown in doc pages

[shreya1510ss]

What has changed?

Every docs page is now also available as plain Markdown, and llms.txt / llms-full.txt are auto-generated at build time — so AI agents (and humans) can grab clean per-page content instead of parsing HTML.

• Added docusaurus-plugin-llms, configured to generate llms.txt, llms-full.txt, and a .md twin for every doc page (same URL, .md extension).
• Added a "Copy as Markdown" / "View as Markdown" control under each page title, so the .md file is actually discoverable, not just guessable. Styled with Tailwind utility classes to match this repo's existing convention.
• Patched docusaurus-plugin-llms (via patch-package, auto-reapplied on install) to fix two real content bugs found during review:
  • Some pages reuse content from another file via MDX imports (e.g. import WhatIsKeploy from '../concepts/what-is-keploy.md'); the plugin only resolved these when the filename had an underscore, so ~7 pages were leaking that raw import/JSX line into their .md output instead of the real text.
  • ~106 pages were also leaking raw React component code (<Tabs>, <ProductTier>, <HowTo>, etc.) into their .md output. Fixed by stripping self-closing component tags entirely and unwrapping paired ones while preserving their real inner content — carefully excluding <YOUR_INGRESS_HOST>-style fill-in-your-own-value placeholders, which look similar but aren't components.
• Fixed a timer leak in the copy button: a setTimeout resetting its label was never cleared, so it could fire after the component unmounted (React warning) or race itself on repeated clicks.

Fixes https://github.com/keploy/enterprise/issues/2272

How Has This Been Tested?

Ran npm run build then npm run serve — confirmed the build succeeds, llms.txt/llms-full.txt generate correctly, each page's .md twin loads at its matching URL, and the Copy/View buttons work in both light and dark mode.

Preview.Video.mov

[github-actions]

Thank you and congratulations 🎉 for opening your very first pull request in keploy

[Copilot]

Pull request overview

Adds first-class “LLM-friendly” outputs for the docs site by generating per-page .md twins plus llms.txt / llms-full.txt, and exposes the per-page Markdown via UI actions under each doc title.

Changes:

• Add and configure docusaurus-plugin-llms to generate llms.txt, llms-full.txt, and per-page Markdown outputs at build time.
• Add a MarkdownPageActions UI control to “Copy as Markdown” / “View as Markdown” for eligible doc pages.
• Wire the new control into the DocItem theme and update lockfile for new dependencies.

Reviewed changes

Copilot reviewed 4 out of 5 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
yarn.lock	Locks new transitive dependencies required by docusaurus-plugin-llms.
package.json	Adds docusaurus-plugin-llms dependency.
docusaurus.config.js	Configures the plugin to generate llms*.txt and Markdown per doc page.
src/theme/DocItem/index.js	Computes .md twin URL and renders Markdown actions under the doc title.
src/components/MarkdownPageActions.js	Implements “Copy as Markdown” / “View as Markdown” UI and clipboard logic.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 5 out of 6 changed files in this pull request and generated no new comments.

[manasmanohar]

Please address the comments below. Specifics left inline; this one isn't tied to a diff line:

Collides with the llms.txt we already ship.
We already have curated static/llms.txt and static/llms-full.txt (content team's — About Keploy, awards, positioning, curated links). The plugin generates the same two filenames, and its postBuild runs after static/ is copied, so it overwrites them: build/llms.txt is the generated version and the curated content is gone. Net effect — the content team's file is silently replaced, and editing static/llms.txt later would have no effect. The static file also drifts today since it's hand-maintained.
Option: merge both — feed the curated text into the plugin via rootContent/fullRootContent (+ title/description) so it stays on top and the auto-generated index below stays fresh. Needs the curated source moved out of static/ (so it isn't overwritten) and the static copies removed. Worth confirming the exact split with the content team.

[manasmanohar]

Copy/View as Markdown 404s on the 5 section landing pages (concepts, dependencies, operation, server, server/sdk-installation). For an index page the permalink is /docs/concepts/, so this produces /docs/concepts.md — but the plugin writes the file to /docs/concepts/index.md, so the link 404s and Copy shows "Copy failed". Normal pages are fine.
Options: special-case folder permalinks (trailing slash → append index.md instead of .md), or derive the .md URL from the plugin's output instead of reconstructing it from the permalink.

[manasmanohar]

These two regexes run over the whole document including fenced code, so real code gets stripped. e.g. installation/windows.md: wsl --install -d <Distribution Name> → wsl --install -d (~5 files); quickstart/java-spring-boot-xml.md: the <User>...</User> wrapper is removed and the XML sample ends up malformed. This is new — those tags were preserved before (the plugin only stripped a fixed HTML allowlist).
Option: skip fenced code blocks when stripping (mask fences → strip prose → restore).

[manasmanohar]

Plugin is ^0.4.0 but the patch is pinned to 0.4.0 — patch-package will fail the install if a 0.4.1 publishes. Consider pinning the exact version.

[manasmanohar]

docsDir is hardcoded to versioned_docs/version-4.0.0 — will need updating when we cut the next docs version, since the buttons show on whatever the latest version is (gated on isLatestVersion).

[manasmanohar]

add fixed width to prevent the div from shrinking when changing text