Does your agent make things up when it can’t quite place a query? Search Fragments stops that.

On the queries a model can’t confidently place, it resolves the real answer, points to the sources, or honestly declines — it never bluffs.

Connect

It’s an MCP server. One tool: resolve_fragment. MCP 2025-11-25 Streamable HTTP. Free, no signup.

https://searchfragments.com/api/mcp

{
  "mcpServers": {
    "search-fragments": {
      "url": "https://searchfragments.com/api/mcp"
    }
  }
}

When an agent hits a half-remembered, cross-source query it can’t pin down, it guesses — confidently, and often wrong. Search Fragments is built for exactly those queries. It gives you a verified answer, a ranked shortlist to piece together, or an honest “not resolvable.” Never a confident-wrong. If your agent has this problem, wire it in and try it — the demo below runs a real resolution right now.

• Resolved A named candidate, with a confidence level.
• Shortlist A ranked list of sources to piece together.
• No resolution An explicit “not resolvable from text.”

Built to decline rather than guess.

Tested

We ran it against a baseline agent on 50 hard, under-documented fragments — the kind a model can’t confidently place. On three, the baseline fabricated a specific answer: it invented a Japanese director for a real film, named a Ronnie Barker mime sketch that was never performed, and attributed a bicycle-and-marriage study to a geneticist who never published it. Search Fragments declined honestly on all three.