Add Kreuzberg Codex plugins

[Goldziher]

Adds three related Kreuzberg plugins to Tools & Integrations:

• Kreuzberg: local document extraction for 91+ formats.
• Kreuzberg Cloud: managed extraction workflows with usage, job tracking, sandbox, and presigned-upload skills.
• Kreuzcrawl: web crawling and scraping workflows for Codex.

This also updates the generator and PR validator to support multiple Codex plugin manifests from a single monorepo by matching README entries to the appropriate nested .codex-plugin/plugin.json root and validating nested bundle directories.

Validation run locally:

• python3 scripts/check-alphabetical.py README.md
• python3 scripts/validate-plugin-pr.py --base-ref origin/main
• python3 -m py_compile scripts/generate_plugins_json.py scripts/validate-plugin-pr.py
• git diff HEAD --check
• uvx plugin-scanner scan plugins/kreuzberg-dev/plugins/plugins/kreuzberg --format text -> 91/100
• uvx plugin-scanner scan plugins/kreuzberg-dev/plugins/plugins/kreuzberg-cloud --format text -> 91/100
• uvx plugin-scanner scan plugins/kreuzberg-dev/plugins/plugins/kreuzcrawl --format text -> 91/100

[gemini-code-assist]

Code Review

This pull request introduces three new plugins (kreuzberg, kreuzberg-cloud, and kreuzcrawl) along with their associated skills, assets, and comprehensive reference documentation. It also updates the marketplace metadata, reorders several plugins alphabetically, and enhances the generator and validator scripts to support monorepos containing multiple plugins. The code review feedback highlights a critical bug in the plugin generator script where sibling or nested plugins in a monorepo could delete each other's mirrored files during shutil.rmtree operations, and identifies a casing mismatch in a TypeScript configuration example within the advanced features documentation.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

In monorepos with multiple plugins, if one plugin is located at the repository root (relative path "") and another is in a subdirectory, clearing the destination_root using shutil.rmtree for the root plugin will delete the subdirectory plugin's mirrored files if the root plugin is processed after the subdirectory plugin. To prevent sibling or nested plugins from deleting each other's files, we should clear the base repository destination directory once per repository instead of clearing the destination_root per plugin.

    repo_root = PLUGINS_ROOT / plugin["owner"] / plugin["repo"]
    import threading
    global _cleared_repos, _cleared_repos_lock
    if "_cleared_repos" not in globals():
        _cleared_repos = set()
        _cleared_repos_lock = threading.Lock()
    owner_repo = f"{plugin['owner']}/{plugin['repo']}"
    with _cleared_repos_lock:
        if owner_repo not in _cleared_repos:
            if repo_root.exists():
                shutil.rmtree(repo_root)
            _cleared_repos.add(owner_repo)
    destination_root = repo_root
    if use_subpath and plugin_root_relative:
        destination_root = destination_root / PurePosixPath(plugin_root_relative)

[gemini-code-assist]

In the TypeScript example for securityLimits, the keys are written in snake_case (e.g., max_file_size), but the Node.js/TypeScript SDK uses camelCase for configuration fields (e.g., maxFileSize). Update these keys to match the SDK's naming conventions.

Suggested change

	```typescript
	import { ExtractionConfig, extractFile } from '@kreuzberg/node';
	const config: ExtractionConfig = {
	securityLimits: {
	maxFileSize: 100_000_000, // 100 MB
	maxArchiveFiles: 1000,
	maxTextLength: 10_000_000, // 10 MB of text
	maxPages: 10000,
	maxConcurrentExtractions: 4