mcp server
io.github.ipezygj/evalgate
Statistical checks an agent runs before trusting an AI eval number (is #1 real, judge bias, more).
Description as published by the maintainer. Source
- version 0.4.1
- active
- evaluation
active — Registry entry last updated 2026-07-23. Dashed tags are derived by ZBS Index from the published description, not stated by the maintainer.
Signals
These are separate measurements of different things. They are deliberately not combined into one score, because a popularity number that mixes website traffic with saves and stars cannot be checked or acted on.
| Signal | Value | What it measures | Window | Observed | Source |
|---|---|---|---|---|---|
| Latest published version | 0.4.1 | Latest version string the maintainer published to the registry. | as of fetch | Model Context Protocol | |
| Registry record last updated | 2026-07-23 | When the registry record was last updated by its maintainer. | point in time | Model Context Protocol | |
| First listed in the MCP Registry | 2026-07-23 | Date this server was first published to the official MCP Registry. Not a usage or quality measure. | point in time | Model Context Protocol |
Where to get it
Also from ipezygj
-
io.github.ipezygj/agent-guard
Safety checks an agent runs before it acts: package malware, destructive command, secret leak.
-
io.github.ipezygj/numguard
Verify a number before an agent asserts it — evals, backtests (Deflated Sharpe), signed receipts.
Bring your own setup
We take apart real AI setups every week and show what broke, what cost too much, and what the trace actually said. If you run agents on real work, that is where the useful conversation is.
Join ZBS AI Practice Lab