Self-Hosted Web Research & Hybrid Search Engine for Local LLMs
WebFind is a self-hosted web research engine and Google-style search interface built in Rust. It crawls, indexes, ranks, and searches the live web for local and free LLMs without API keys, cloud subscriptions, or auth barriers, persisting link graphs to an embedded Turso store.
Critical friction and failure modes observed in existing workflows
Free and locally hosted LLMs (Ollama, LM Studio, vLLM) lack real-time web search capabilities, which commercial providers lock behind expensive subscription tiers.
Engineering methodology, model selection, and pipeline design
Engineered a single-binary search system featuring a high-concurrency crawler with headless Chromium (CDP) support, dense vector embedding ranking (fastembed-rs) fused with Tantivy BM25 and PageRank via Reciprocal Rank Fusion, all persisted in an embedded Turso/libSQL database.
Validated benchmarks, latency figures, and operational efficiency
Provides 100% private, zero-cost, real-time web research to autonomous AI agents and humans through both a single-call CLI and a real-time SSE Web UI.
Single-call CLI research command crawling, persisting, and ranking live web pages
Hybrid BM25 + dense vector semantic search fused via Reciprocal Rank Fusion (RRF)
Embedded Turso/libSQL graph memory accumulating link graphs and PageRank over time
Deep crawler with headless Chromium (CDP) for JavaScript-rendered SPA extraction
Google-style responsive Web UI with instant autocomplete and SSE live streaming
Clean agent-native noise-free markdown extraction stripping ads and boilerplate
Production libraries, architectural components, and runtimes used in this system:
Let's discuss architecture, production deployment, or building high-performance intelligence solutions.