Every premium and members-only resource for $9.99/month, as a no-commitment subscription. See the plans

PalateKit

Menu

Request a resource

Describe the resource you are missing in one sentence. We will email you when it is published.

0 / 280

Unlock PalateKit

Unlock everything

Plan

Paying means accepting our terms of sale.

Secure payment by Stripe. Receipts and invoices sent by email.

Welcome back

Sign in to install, download or save this resource. A free account is enough for free resources.

Auditing an evaluation or metric for leakage v1.0.0

Use this resource to check that an evaluation is not just internal confirmation before trusting it

New

  • 300 downloads
  • 149 likes
  • 197 saves

Free

Install with the CLI

In a terminal, from the folder you want:

npx --yes palatekit@latest install validation-leakage-audit

With no option, the skill goes to ~/.agents/skills/; add --agent claude-code (or codex, cursor…) to target an agent.

First time: connect the CLI with

npx --yes palatekit@latest login

The agent identifies the validation, names its components (model, scorer, designer) and keeps only the leakage patterns that apply, each with an independence fix. It changes nothing.

Mostly picked by

Updated November 2, 2026

Card for the “Auditing an evaluation or metric for leakage” skill

Details

Generates with
Your agent's model
Licence
MIT

Similar resources

View all

What you get

  • SKILL.md: step-by-step audit method and read-only rules
  • SKILL.md: the eight leakage patterns and their fixes

Good fit for

  • A model score or A/B result looks too clean
  • You are preparing a validation set and want to check its independence

Not for

    Tags

    • evaluation
    • data leakage
    • validation

    About

    You say

    “Audit our quality metric: does the same model both generate and grade the answers?”

    Card for the “Auditing an evaluation or metric for leakage” skill

    Skill published by lilmgenius under the MIT license, delivered in its original version.

    Show more

    How to use it

    1. Install the skill in your agent's skills folder
    2. Ask your agent to check that an evaluation is not just internal confirmation before trusting it
    3. Review and adjust the result

    Content preview

    Files

    • LICENSE1.1 KB
    • NOTICE-PALATEKIT.md0.5 KB
    • SKILL.md3.2 KB

    Origin and licence

    Listed unmodified

    Author
    lilmgenius
    Source
    github.com/lilmgenius/paperthin/tree/7d5dc6235990230e45a24a25c527d81701da5458/skills/depth/mandela
    Licence
    MIT

    Notice

    « mandela » (lilmgenius, MIT) livré dans sa version d'origine : fiche en français et en anglais par PalateKit.

    User reviews

    –

    No ratings yet

    Loading reviews…

    Delete your review?

    Your rating and comment will be removed from the page.