Local-based Web Search MCP Server 'wigolo' Released for AI Agents

Developed to solve the web access cost problem of AI coding agents, wigolo is an MCP (Model Context Protocol) server that runs search engines and collection engines on the user's local machine.

Key Features and Functions:

• Zero Cost: Core tools (search, fetch, crawl, extract, etc.) operate locally without an API key, incurring no per-query cost.
• 10 Specialized Tools Provided:
  • search: Performs multi-engine search and ML reranking
  • fetch: Page acquisition and markdown conversion using a headless browser
  • crawl: Multi-page crawling and robots.txt compliance
  • extract: Extraction of structured data (JSON-LD, schema, etc.)
  • research & agent: Autonomous collection loops and report synthesis
• Data Security: All data is processed within the local environment (~/.wigolo/) to prevent external leakage.
• Flexible LLM Integration: LLM is used only optionally when synthesizing research results, and can either remain fully local via Ollama or optionally use external APIs such as Gemini and OpenAI.
• Explainable Results: Provides engine telemetry, score breakdowns, citation IDs, etc. for search results, helping agents make evidence-based judgments.