Three dominant themes from the discussion
| # | Theme | Key takeaway (with supporting quote) |
|---|---|---|
| 1 | Self‑hosted meta‑search (SearXNG) is the preferred solution for privacy‑conscious users and LLM tool‑calling | > "I've been self‑hosting this as my default engine across all of my searches for a few years now. I can't recommend it more highly." – drnick1 > "TinySearch wraps this and works well for agents… it optimizes the context before it even gets to the agent so as to not waste tokens." – satvikpendem |
| 2 | There is a strong need for an MCP‑compatible, JSON‑API‑enabled search service that can be directly called by local models | > "It appears to be a key tool for providing search to local models." – goodroot > "Enable the JSON API ( search.formats: [html, json]; off by default)." – configuration note |
| 3 | Alternative projects (e.g., Hister) that index personal browsing history and the question of scalability are widely discussed | > "My long‑term goal is exactly this. I'd like to add federation/P2P feature to evolve from being a private search companion." – asciimoo > "There has been no stress tests in this regard. The indexer lib Bleve can handle millions of documents." – asciimoo |
Summary
- Privacy‑first self‑hosting – Users favor SearXNG for full control and integration with LLMs.
- Tool‑callable search – The community emphasizes an MCP/JSON API to let local models invoke search without wasting tokens.
- Local indexing & scalability concerns – Projects like Hister are explored as alternatives, with ongoing debate about handling massive data volumes.