Download and normalize official documentation pages into local .mdx files at user-specified paths using markdown.new. Use when user asks to fetch docs from a...
--- name: official-docs-to-mdx description: Download and normalize official documentation pages into local .mdx files at user-specified paths using markdown.new. Use when user asks to fetch docs from a URL, archive official docs locally, sync docs snapshots, or convert docs pages into clean MDX with frontmatter (title, description, sourceUrl, retrievedAt). --- # Official Docs to MDX Convert a docs URL into a cleaned local `.mdx` file at a path the user specifies. Use this skill when the user wants reproducible, file-based docs snapshots for local knowledge bases, RAG prep, or project docs folders. ## Quick Start Run: ```bash scripts/mdnew_to_mdx.sh <source_url> <output_mdx_path> ``` Example: ```bash scripts/mdnew_to_mdx.sh \ https://docs.convex.dev/agents/agent-usage \ docs/convex/agent-usage.mdx ``` ## Workflow 1. Validate inputs. - Require exactly 2 args: source URL and output path. - Keep URL as full `https://...` form. 2. Convert with script. - Execute `scripts/mdnew_to_mdx.sh`. 3. Confirm output. - Verify file exists. - Print or inspect frontmatter for `title`, `description`, `sourceUrl`, `retrievedAt`. 4. Add indexes for folder-based docs (required). - Any docs folder must include `index.mdx`. - If `docs/` exists, `docs/index.mdx` must exist. - Index files should link child docs/subfolders. - Keep frontmatter: `title`, `description`, `sourceUrl` (`local://...` allowed), `retrievedAt`. 5. Report result. - Return output path(s), overwrite status, and any created/updated index files. ## Output Contract The generated MDX must include frontmatter: - `title` - `description` - `sourceUrl` - `retrievedAt` (UTC `YYYY-MM-DD`) Body is cleaned to remove common docs-site artifacts and excess blank lines. ## Script Details For behavior details and cleanup rules, read: - `references/mdnew_to_mdx.md` Use the bundled script directly: - `scripts/mdnew_to_mdx.sh` ## Notes - Dependencies: `bash`, `curl`, `awk`, `mktemp`. - Output file is overwritten if it already exists. - Re-run anytime to refresh docs snapshot from upstream.
don't have the plugin yet? install it then click "run inline in claude" again.