Show HN: GoogLLM – Google search that returns Markdown instead of HTML https://ift.tt/Y2NJZ3e

Show HN: GoogLLM – Google search that returns Markdown instead of HTML As part of my bigger goal to make the web more agent-friendly, this weekend i decided to tackle google. The "AI-native" search APIs like Tavily and Exa exist, but they require setup and don't actually use Google's results. So I built something simple - a proxy that takes Google search URLs and returns the results as clean markdown instead of HTML. You literally just change "google.com" to "googllm.com" in any search URL. ```bash # Returns 500KB of HTML: curl " https://ift.tt/9mkJwB0 " # Returns clean markdown: curl " https://ift.tt/K7pD4ei " ``` *What it does:* - Serves normal HTML to browsers (so humans can use it normally) - Returns markdown to everything else (curl, fetch, LLM agents) - Supports all Google search types: web, images, news, scholar, shopping, etc. - No auth needed for testing (10 requests/hour free) *Technical approach:* - Content negotiation based on Accept headers - Caches results to avoid hammering Google - Simple pricing: 0.5¢ per search after free tier I built this over a long weekend because I was tired of writing HTML parsers for every project. The whole thing is designed around a single principle: make Google search results consumable by LLMs without any complexity. *Questions for HN:* - Is this approach too simplistic? Should search APIs be more complex? - How do you currently handle search in your LLM applications? - Any concerns about the proxy approach vs. building from scratch? The llms.txt documentation is intentionally comprehensive (2500 tokens) so any LLM can understand and use it immediately. Live demo: https://googllm.com API docs: https://ift.tt/UeHYpdo Would love feedback on the approach and any edge cases I might have missed. https://googllm.com June 3, 2025 at 12:37AM