You rarely ask for this one directly. When you hand your assistant a link, a username, a title or a person’s name, it uses this tool to find the matching UpTrust account or post, so it can check or rank it. Finding something says nothing about whether to trust it. If more than one account matches a name, your assistant should ask which one you meant. If nothing matches, UpTrust has no record of it.
Ask it like this
- “Should I trust this article? (with a link)”
- “Is this the Sam Lee who writes about housing?”
When another tool fits better
- Can I trust this source?
You want to know whether to rely on it. This tool only finds it.
Before you use it
- It needs the Read your trust signal permission, which UpTrust asks you to approve when you connect.
- It only reads. It changes nothing on UpTrust.
- In Claude, connect
https://mcp.uptrusting.com/mcp. In ChatGPT, connecthttps://mcp.uptrusting.com/chatgpt. How to connect. - It can also answer from another trust graph than yours, such as one built from academic papers; see which communities it can answer from.
What your assistant is told
Your assistant decides when to use this tool from the title and description UpTrust’s server gives it. Here they are exactly as the server sends them.
- Tool name
resolve_entity- Title
- Turn a URL, name, or title into an ID you can check
- Description
Use this when the user asks about a specific source, person, post, paper or site by its link or name ("should I trust this source?", "can I trust this source?", "is this site legit?", "is this the right person?") and you need the ID the other UpTrust tools take. Pass what they hold — a URL, a username or handle, a paper or post title, a repo or package name, a person's name. It returns the matching entities, each with an ID, a `kind` (`user` or `post`), a display name, a snippet and a lexical `match` band. Every other trust tool takes IDs in this issuer's namespace, and this is the tool that produces them: call it first whenever the user hands you a link or a name rather than an ID, then pass the ID with its `kind` to is_trusted_by_principal (one item: can they rely on it?) or rank_for_principal (several items to order). It does not judge trust: `match` is lexical confidence only and carries NO trust signal — resolving something says nothing about whether it is trustworthy. MORE THAN ONE CANDIDATE IS A NORMAL ANSWER: ask the user which they meant instead of guessing. AN EMPTY LIST IS ALSO A NORMAL ANSWER: it means this graph has no such entity, which is worth saying plainly rather than retrying. BE SLOWEST TO PICK WHEN THE CANDIDATES ARE PEOPLE. A personal name is evidence of identity, never proof of it, and distinct accounts routinely answer to the same name. Asking the user which one costs a turn; acting on the wrong person does not undo. Read `is_agent` before you present anyone: an agent account is not a human, and on a disambiguation surface that is the most decisive thing you can say about two similar-looking candidates. Cheap, generously rate-limited, and safe to call speculatively: it reads no per-person trust graph and notifies nobody.