What it does
This MCP server crawls deepwiki.com URLs, extracts HTML, converts it to Markdown, and returns content either aggregated into a single document or as structured per-page objects. It sanitizes HTML by removing navigation, headers, footers, and ads, then rewrites links for Markdown compatibility. Crawl depth and concurrency are configurable via environment variables.
Who it's for
Backend engineers and researchers who need to ingest DeepWiki documentation into Claude or other models as Markdown. Teams maintaining local knowledge bases of library documentation (shadcn, Tailwind, etc.) would benefit—if the server were functional.
Common use cases
- Extract multi-page library guides into a single Markdown file for inclusion in LLM context windows
- Batch-convert DeepWiki documentation pages into structured Markdown objects for offline processing
- Build training datasets from community-authored library documentation
Setup pitfalls
- Non-functional: DeepWiki has blocked automated scraping. The server cannot retrieve content and its README recommends using the official DeepWiki MCP instead.
- Maintenance is stale (last commit 77 days ago, CI failing). Do not expect updates or fixes.
- Configuration via
DEEPWIKI_MAX_CONCURRENCY,DEEPWIKI_REQUEST_TIMEOUT,DEEPWIKI_MAX_RETRIES, andDEEPWIKI_RETRY_DELAYenvironment variables, but these cannot resolve the upstream block.