Your assistant gets the communities that pull on your judgment: ones you belong to, and ones whose influence reaches you through people you trust. It also gets where you sit among them and the items several communities agree on, with live divides marked. Everything is banded or counted, never a raw score. On a genuinely contested topic, “no common ground” is a real answer.
Ask it like this
- “Which camps shape how I see this?”
- “What is my epistemic neighborhood on economics?”
When another tool fits better
- Which of these should I trust first?
You want a list put in order. - Can I trust this source?
You want one item checked. - Who should I meet?
You want someone to meet.
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
field_for_principal- Title
- Why do my rankings look this way? (which camps pull on them)
- Description
Use this when the user asks why they see or rank things the way they do: "which camps shape how I see this?", "why do I rank these this way?", "what's my epistemic neighborhood?". It returns the named communities that pull on their judgment, each pull marked "claimed" (their own community) or "propagated" (gravity from communities they never joined, arriving through the people they trust); where the user actually sits (a soft membership vector, never one label); and the load-bearing items endorsed ACROSS communities' own readings, with contested_between marking live divides. It describes the landscape and calibrates deference; it does not rank a list (rank_for_principal), score an entity (is_trusted_by_principal), or introduce anyone (who_should_principal_meet). Everything is banded or counted, never a raw score; flat_space=true ("genuinely contested, no cross-community convergence") is a first-class answer, not a failure. Response schema is versioned; a backend 404 means this issuer has not shipped the tool, not an auth problem.