Skip to content

[New Server] Dominion Observatory — runtime behavioral trust scoring for MCP servers #4198

@vdineshk

Description

@vdineshk

Dominion Observatory — MCP server for runtime behavioral trust scoring

Proposing the addition of the Dominion Observatory as a reference MCP server in this collection.

What it does

The Observatory is a behavioral trust registry that tracks runtime telemetry for MCP servers — latency, success rates, anomaly detection, trust scores — and exposes this data via MCP tools so agents can make trust-informed decisions about which servers to call.

MCP tools exposed

Tool Description
check_trust Query a server's current behavioral trust score (0–100) + anomaly flags
report_interaction Report anonymized telemetry after a tool call (server_url, success, latency_ms)
get_leaderboard Top servers by category, ranked by behavioral trust
get_baselines Statistical baselines (avg latency, p95, success rate) for comparison
check_anomaly Check if a server is currently exhibiting anomalous behavior
register_server Register a new MCP server for tracking
get_server_history Historical trust score trend for a server
observatory_stats Overall Observatory statistics
get_compliance_report Compliance audit report for a server over a date range

Why this belongs in the reference servers collection

  1. Meta-MCP: This is an MCP server that scores other MCP servers. As the ecosystem grows (14,800+ servers tracked currently), runtime behavioral trust becomes a necessary infrastructure layer.
  2. Agent safety: Agents calling unknown MCP servers have no way to evaluate runtime reliability. Static allow-lists don't catch behavioral drift. The Observatory fills this gap.
  3. Privacy-preserving: Only anonymized telemetry (server_url, success, latency_ms, tool_name, http_status). No prompts, tool arguments, outputs, user IDs, or IPs.
  4. Already live: The Observatory is running on Cloudflare Workers with D1, tracking 14,800+ servers and 87,000+ interactions. Not a proposal — working infrastructure.

Technical details

  • Endpoint: https://dominion-observatory.sgdata.workers.dev/mcp
  • Transport: Streamable HTTP (MCP spec compliant)
  • Infrastructure: Cloudflare Workers + D1
  • Source: github.com/vdineshk/daee-engine
  • Also available as: REST API, /llms.txt, /benchmark/<server-name> endpoint

SDK support

  • Python: pip install dominion-observatory — includes LangChain callback handler
  • TypeScript: @dominion/trust-provider on npm — includes x402 beforeSettle hook

Integration with the broader ecosystem

Happy to submit a PR adding the Observatory to the servers collection if there's interest. The server implementation is self-contained and follows the MCP transport spec.

References

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions