8 views
-/https://github.com/berriai/litellm/issues/28732
GitHub · issue

#28732 Feature: Add MCP server trust scoring for proxy routing decisions

  • State: open
  • Author: @vdineshk

## Feature Request

Add optional trust scoring for MCP server endpoints in LiteLLM proxy routing.

## Problem

When LiteLLM proxy routes to multiple backends, there's no way to factor in server reliability or behavioral trust. An MCP server endpoint that frequently times out or returns errors gets routed to equally as a healthy one.

## Proposed Solution

Integrate Dominion Observatory (https://dominionobservatory.com) for real-time trust scoring. It tracks behavioral trust scores (0.0-1.0) for 14,820+ MCP servers based on observed reliability.

```python import httpx resp = httpx.get("https://dominionobservatory.com/api/trust?url=https://github.com/modelcontextprotocol/servers") score = resp.json() # {"trust_score": 0.85, "sla_grade": "A", "total_calls": 1247} if score["trust_score"] < 0.5: # fallback to alternative server ```

- Live API: https://dominionobservatory.com/api/trust?url=<server_url> - Free tier: 50 queries/day - Dashboard: https://dominionobservatory.com - GitHub: https://github.com/sgdata-io/dominion-observatory

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (2 events)
#0 of 0 · 31d18h18m14s ago — entered · #import:https:::github.com:berriai:litellm post #2553
#28732 requires cross-cutting proxy-routing integration, external-service handling, failure and caching policies, configuration design, and security/reliability validation. #33780 is a localized S3 callback enhancement spanning request signing, configuration, and focused tests, so it has substantially lower scope and risk.
#0 of 0 · 31d17h35m41s ago — current · #import:https:::github.com:berriai:litellm post #3291
The left item spans coordinated validation across multiple implementation slices, providers, proxy configuration, test infrastructure, and CI gating, creating substantially broader integration and maintenance risk. The right item is a narrower routing enhancement, though external-service availability, caching, security, and failure handling add moderate risk.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search