iris-eval/mcp-server

MCP-native agent evaluation and observability server

46
/ 100
Emerging

Provides hierarchical trace logging with per-tool latency and cost tracking stored in SQLite, plus 12 built-in evaluation rules (PII detection, hallucination markers, prompt injection patterns) and custom rule support via Zod schemas. Exposes three MCP tools (`log_trace`, `evaluate_output`, `get_traces`) that any MCP-compatible agent discovers automatically via stdio transport, with optional web dashboard and HTTP mode for team deployments. Integrates with Claude Desktop, Cursor, Windsurf, LangChain, and CrewAI.

Available on npm.

Maintenance 13 / 25
Adoption 3 / 25
Maturity 18 / 25
Community 12 / 25

How are scores calculated?

Stars

3

Forks

1

Language

TypeScript

License

MIT

Last pushed

Mar 18, 2026

Commits (30d)

0

Dependencies

8

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/mcp/iris-eval/mcp-server"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.