DevLens MCP

Instructions

Implementation Reference

• src/devlens/tools/advanced.py:106-156 (handler)

  The core implementation of the logic for finding related pages. It fetches content from the provided URL, extracts a search query, performs a web search, and returns a formatted report of related URLs.

  async def find_related(url: str, limit: int = 5) -> str:
      """Find related pages to a given URL.
  
      Args:
          url: Base URL to find related content for.
          limit: Maximum related pages (1-10).
  
      Returns:
          List of related pages with descriptions.
  
      Example:
          >>> related = await find_related("https://docs.python.org/3/library/asyncio.html")
      """
      limit = min(max(limit, 1), 10)
  
      # Extract topic from URL
      try:
          doc = await _scraper.fetch(url, retry=1)
          # Use title as search query
          search_query = f"{doc.title} related documentation"
      except Exception:
          # Fallback to URL-based query
          parsed = urlparse(url)
          path_parts = parsed.path.strip("/").split("/")
          search_query = " ".join(path_parts[-2:] if len(path_parts) > 1 else path_parts)
  
      # Search for related content
      try:
          results = await _ddg.search(search_query, limit=limit + 5)
      except SearchError:
          return f"# Related Pages\n\nFailed to find related content for: {url}"
  
      # Filter out the original URL
      related = [r for r in results if r.url != url][:limit]
  
      if not related:
          return f"# Related Pages\n\nNo related pages found for: {url}"
  
      # Build report
      report_lines = [
          "# Related Pages\n",
          f"> Based on: {url}\n",
          "## Recommendations\n",
      ]
  
      for i, r in enumerate(related, 1):
          report_lines.append(f"\n### {i}. {r.title}\n")
          report_lines.append(f"**URL**: {r.url}\n")
          report_lines.append(f"{r.snippet}\n")
  
      return "\n".join(report_lines)

• src/devlens/server.py:131-143 (handler)

  The MCP-exposed wrapper function for the `tool_find_related` tool, which calls the underlying `find_related` implementation.

  async def tool_find_related(url: str, limit: int = 5) -> str:
      """Find pages related to a given URL.
  
      Uses the page content to discover similar resources.
  
      Args:
          url: Base URL to find related content for.
          limit: Max related pages (1-10, default 5).
  
      Returns:
          List of related pages with descriptions.
      """
      return await find_related(url, limit)